Document Type: 
Document Number (FOIA) /ESDN (CREST): 
Release Decision: 
Original Classification: 
Document Page Count: 
Document Creation Date: 
November 1, 2016
Sequence Number: 
Case Number: 
Content Type: 
PDF icon CIA-RDP82-00850R000400020012-2.pdf8.34 MB
APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIbL USE ONLY JPRS L/9773 5 June 1981 U~SR R~ ort ~ _ CYBERNETICS, COMPUTERS AND AUTOMATION TECHNOLOGY CFOt~O 15/81) FBIS FOREIGN BROADCAST IP~FORMATION SERVICE FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 NOTE JPRS publications contain information primarily from foreign - newspapers, periodicals and books, but also from news agency transmissions and broadcasts. Materials from foreign-language sources are translated; th.ose from English-language sources are transcribed or reprinted, with the original phrasing and other characteristics retained. ' Headlines, editorial reports, and material enclosed in brackets are supplied by JPRS. Processing indicai:ars such as [TextJ - or [Excerpt] in the first line of each item, or following the last line of a brief, indicate how the original information was processed. W'here no processing indicator is given, the infor- mation was summarized or extracted. Unfamiliar names rendered phonetically transliterated are enclosed in parentheses. Words or names preceded by a ques- tion mark and enclosed in parentheses were not clear in the original but have been supplied as appropriate in context. ~ Other unattributed parenthetical notes within the body of an _ item origina~e with the source. Times within items are as given by source. The contents of this publication in no way represent the poli- - cies, views or at.titudes of the U.S. Government. COPYRIGHT LAWS AND REGULATIONS GOVERNING OWNERSHIP OF - MATERIALS REPRODUCED HEREIN REQUIRE THAT DISSEMINATION _ OF THIS PUBLICATION BE RESTRICTED FOR OFFICIAL USE ODTL,Y. = r ~ APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFF:CIAI. USE ONLY JPRS L/9773 ~ 5 June 1981 USSR REPORT CYBERNETICS, COMPUTERS AND AUTOMATION TECHNOLOGY cFOUO is/8i) _ CONTENTS xaxn~~rax~ Hybrid LSI Manufacturing Technology 1 Characteristics of Manufacturing Printsd-Cir.r,uit Cards Usi:~g SPF-2 Dry Film Photoresist 3 ~ Multimachine and Multiprocessor Systems.......~ 9 Basic Characteristics of Some Domestic LSI Microprocessor Complexes 11 P320 Programmable Calibrator 12 _ ilew Computer Hardware............................................. 14 S OFTtidARE System for Complex Debugging of Elektronika S5-12 Microcomputer Based on Functional MicroprocesQor Modules 18 Main Trends in Form~tion of an Algorithm and Program Fund for OASU 22 Comparative Course of the PL/1 Language (on the Basis of ~ ALGOL-60) 24 Using Data Bases in the ASU TP of Large-Capacity Units............ 27 MUI,TIPROCESSOR SYSTEMS - State Commission Accepts Institute of Cybernetics Design for ~ Multiprocessor Supercomputer 33 Construction of Algorithmic Language Families for Programming and Designing Multiprocessor Comput~r Systems 34 _ ~ a- [III - USSR - 21.C S&T FOUO] r. nut nr, r,Kr+r . t r ror, nwrY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY VOICE RECOGNITION/SYNTHESIS Mini-Computer Voice C~unication System 45 ~ NETWORKS Concept of 3 Modern Computing Network 50 ~ Data Transmission System (SPD) of the 'E1'brus-1' T~ultiprocessor Computing Complex (MVK) 67 New Book Discusses Control Systems for Communications Networks.... 88 Hybrid Switching Syste~s 90 - A Methodology for the Simulation Modeling of the ~ollective-Use Computer Center of the Siberian Deparrment of the USSR Academy of Sciences 94 Traffic Distribution Algorithm for Automafied Planning System...... 109 PUBLICATIONS Abstracts From the Journal 'AUTOMATION AND COMPUTER TEC~~IOLOGY'... 110 Table of Contents From ~he Journal 'CYBERNETICS' 114 Cybernetics and Computer Engineering: Digital Control Systems.... 116 Tabl.e of Contents From the Journal 'AUTOMATION AND COMPUTER TECHNOLOGY' 123 - `INFORMATION SELECTION AND TRANSMISSION' No 61 125 'INFORMATION SELECTION AND TRANSMISSION' No 62 127 New Book on Computer Modeling of Filtration in Water Engineering Works 129 ERRATUM: In JPRS L/9766, 1 June 1981 (FOUO 14/81) of this series, _ please change "mobile robot" to "undeYwater robot" in the following placess page 77, next to last line; page 79, caption for fig. 1.7; page 80, eighth line from bottom; page 81, caption for fig. 2.4. - b - FOR OFF~CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY ' HARDWARE I~YBRID LSI MANUFACi'URING TECI~TOLOGY - Moscow PRIBORY I SISTEMY UPRAVLIIJIYA in Russian No 1~ Jan 81 p 40 [Articl.e by engineers A.I. Yakubinskaya~ A.A. Kisilev and I.N. Pervunitskiy: " Features of Manufacturing BGIS (Hybrid Large-Scale Integrated Circuits) Designed for Special Applications"] ~ [Text] The manufacture of hybrid special application LSI with high levels of integration is dictated by their use in digital-analog circuits. The design and technology of manuf~cturing LSI provides for the insta7lation on plates of type K10-17 mounted capacitors~ powerflzl transistors, and no-rack semiconductor circua.ts with flexi.ble and ball-type leads. A complex intracircuit commutation is required here, which is executed in two layers~ separated by an insul.ating layer of orgariic clielectric. Exi.sting manufacturing methods for multilayer separation o~ thin-f`ilm circuits in- volve positioning the f`irst electrical conducting layer on the insulating base~ then a second insulating layer and a second conducting 1~}rer. In opder to achieve electrical contact be~ween the layers~ a selective etching of the dielectric is accomplished using a protective maslz~ a proceas which sometimea causes shorta to occur through the punctures in the protective film~ reaulting in a subsequent reduction in reliability. ~ feature of the proposed technology is the absence of etc, the dielectric film, a conti.nuoua layer 15-20 micrometers in thickness~ a�eature which insures reli- ~.bility in thP entire circuit. ilectrical contact between the metel plated layera is achieved by the uae of inter- layer contacts--~rolume copper specialized leads~ obtained by galuanic buildup on first layer conductors through a positive mask. Copper with a sub-layer of vanadium is employed for the first l~yer conductors. Current aupply is provided through jumpers~ which are then removed. A feature of ',ne plating process is that to obtain equal rate of coating or precipitation upon tti~ entire surface, the base is suspendedin the galvanic bath at an angle of 80 to ~5 degrees to the anode. This process is intensified through the use of variable polarity current. The approaches outlined for the formation of copper columns (leads) 15-20 micrometers in height~ with a height variance on the entire surface of not more than 2 micrometers~ with smooth surface~ fine-crystalline 1 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-00850R040440020012-2 I~OR OFFI~C'lAL L)SF, ONLY structure~ an3 strong adhesion to the sprayed copper. The material used for the interlayer in~ulation is PAK-1~ a poly~.mide electrical insulating lacquer~ which is advantageoua for its high mechanical and electrical inaulating properties over a broad temperature range (from minus 200 to plus 400 degrees C)~ properties 1.ack- _ in~ in o~;her dielectrics. The insulating layer is a continuous film of lacquer i5 . to 20 micormeters thick~ consisting of four layers~ providing high electrophysical parazneters for crossovers, and minimal porosity. The lacquer is poured onto t,he - base surface and is subsequently equally distributed over the entire aurface of the base by centrifuging. To achieve electrical contact of the conductors of the second 1~,}rer with the interlayer jwzctions~ the la~qusr layer located at the tops of the interlayer junctions is removed until th~ tops of the 3unetions are uncovered. For thia~ burnishing or undercutting of the uppper insulating layer m~y be employed. The selection of one method or other is determined by the relief of the film surface . obt~u.ned, which is dependent upon the initial viscosity of the lacquer. With high initiAl vi scosity of the lacquer (approadmately 1500 aec according to the VZ-1~ viscosimeter)~ a layer of polyimide film of vlrtually the same thiclrness as the insulation layer forms on the 3.nterlayer junctions over the entire surface of the base. The f`ilm so formed is suitable for burnishing re~uirementa~ but is not suitable for undercutting. IIse of a lacquer r~rlth initial vlscoaity of 300-l+00 sec produces a film ~,rlth indiatinctly expressed relief~ which ia more suitable for etcPiing. Al1 operations of the technological cycle were performed on series produced equip- raent utilizing existing procedures and operational modes. Thus the ~;echnological process for the manufacturing of DGIS for apecial applica- tions includes aequentiaLly executed operations of vacuum dusting metal plating of the first layer with f~rther formation of the drawing using the method of photo- lithography, formation of the interlayer junctions on conductora of the f5.rst layer through a mask of positive photore~sist uaing jumpers~ removal of the jumpers~ formation of the insulating layer~ uncovering the interlayer junctions~ and the formation of the second layer conductors. Tes~:s of 500 plates manufactured according to the technoloey outlined demonstrated excellent electrophysical paremeters for interlayer insulation: sh rta were lacking for pratical purposes~ and specific capacity was rated at 2-3 pF/mm~. - COP�RIGHT: Izdatel~ stvo ~'Maahinoatroyeniye~~. ~'Pribory i aistemy upravleniya"~ 1981 885t CS 0: 1863 /142 ~ 2 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 ~ FOR OFF'ICIAL USE ONLY UDC 621.38:776.002.5 - CHARACTERISTICS OF P4ANUFACTURING PRINTED-CIRCUIT CARDS USING SPF-2 DRY FILM PHOTORESIST Moscow PRIBORY I SISTEMY UPRAVLENIYA in Russian No 3, May 81 pp 41-42 [Article by engineers M. G. Miko and T. M. Marenkova and Candidate of T~chnical Sciences B. S. Irugov) [TextJ Intensive development of modern computer equipment and instrument building requires solution of a number of complex production problems and primarily develop- ment and serial assimilation of technological processes in producing precision and large printed-circuit cards. Specifically, cards with dimensions of more than 200 mm in which printed-circuit conductors and gaps 150-200 microns wide must be reli- ably formed are used in minicomputers of the SM EVM [International small computer system] and ASVT PS [expansion unkx~ownJ. Many years of experimental investigations conducted at the Institute of Control - Problems (Moscow) and at a nwmber of other Qrganizations show that one of the most effective methods of solving this problem is to use SPF-2 dry film photoresist of type 2-40 or 2-60 and the corresponding highly productive production equipment. Main Properties of SPF-2 Photoresist c Dry film photoresist (SPF) was especially developed to produce printed-circuit cards. It makes it possible to produce conductors with clear edges 150 microns wide and with the same gaps between them, whereas the minimum width of conductors and the gaps between them now prQduced comprises 250-380 microns for liquid photo- resists. Investigations are now being carried out to improve the composition of the photopolymerized composition for SPF-2 photoresist, which will provide a fur- ~ ther increase of its resolution. Since SPF is much thicker than layers of liquid photoresist, electrochemical pre- cipitation of the metal occurs inside the formed channel. For this reason there is no lateral fungus-like growth of tha metal above the edges of the photoresist and consequently one can manufacture cards with more rigid tolerances on the width of conductors and gaps. By using SPF-2 photoresist, one can produce high-quality two-sided printed-circuit cards by the positive image method with pre-drilling of holes with conductor width of 200 microns and gaps between them of 150 microns. The use af SPF also permits 3 _ FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-00850R040440020012-2 FOR OFFICIAL USE ONLY extensive introduction of the "screen" production process when the conductors and the copper in the holes are protected by photoresist during etching. However, this process is applicable only when the contact areas on the phototemplates are not cut. _ The use of SPF-2 photoresist simplifies the production process (drying, tanning and retouching operations are eliminated), complete automation of the process becomes possible, it becomes more productive and permits a sharp increase of card quality and the percentage of yield of acceptable cards. Due to the high chemical stability of SPF-2 photoresis~ in acid and alkaline solu- tions, there are no restrictions on the use of different galvanic baths and etch- - ing agents. Thus, the use of the same alkaline etching solution based on copper chlorate is possible for etching on SPF-2 photoresist (the "screen" method) and on m~tallic resist (tin-lead alloy) . It is important that the protective SPF now being widely developed to produce ther- moresistant electric insulating coatings of printed-circuit cards are also develop- ed in methylchloroform [12]. However, it should be noted that, as shown by operat- ing experience, failure to observe the necessary requirements when ~ising SPF significantly reduces it~ advantages over liquid photoresists. Requirements on Production Buildings , Since SPF is protected by a lavsan film prior to developing, it is less sensitive to dust than liquid photoresists. However, to completely el iminate retouching, the entire process using SPF must be cazried out under "clean r~om" conditions: the incoming air should be filtered and excess pressure should be maintained in . the room. The temperature should be in the range of 20-25�C and relativ~ humidity should com- prise 50 + 10 percent in the section where operations o+ Lamination and exposure of the photoresist are being carried out. Lamination, exp:~ure and development should be carried out in yellow light to prevent prelint~~ar~� polymerization. ~ Requirements on Phototemplates The large thickness of the photopolymerized compasite i.~ t'~~e ~resence of a lavsan film during exposure place more rigid requirements on ~''~ototemplates. The optical density of the opaque sections of the image should not e be low three units of density (otherwise partial polymerization on the prot~ ~~~d sections occurs while that on the transparent sections should not be mor~ ~.r.~.r. 0.1 units of density (other~vise insufficient polymerization of the exg :tions is observed) . A clear, contrast image on the phototemplates is ~a~;c.~i;xed and a corona and washout of lines are not permitted. ' The s lides should have a dark mask around t�z p~~=:;1?" ^4~. Y' of the phototemplate (the minimum distance from the circuit pattern t~~ ~:ac~ ina.y:r, is not less than 10 mm) , - which creates the best conditions during de .�.c~~~ar~t and contributes to more uni- form galvanic buildup of the metal. 4 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFtCIAL USE ONLY ' Centering points in the middle of the contact surfaces are not permitted in the phototemplates. The diameters of the contact surfaces should be 1.5-fold larger than the hole diameter to protect the metallized holes during the "screen" method. Surface Preparation Blanks with pre-drilled holes and which have undergone chemical metallization of the holes with galvanic prolongation by copper to a thickness of 5 microns prior to lamination should be dried in a drying cabinet with air recirculation for 20-30 minutes at t= SO-100�C to remove moisture from the holes. The blanks with the . qalvanic coating are subjected to additional mechanical cleaning. It is necessary that the blanks have an even surface without deep scratches and _ pits, that the edges be even without outgrowths and that no drilling burrs be in - the holes (drilling is accomplished without counterboring). After cleaning, the copper surface should be slightly mat. The best adhesion of ~ the SPF to it is provided by mechanical cleaning, but the prepared surface should not be polished. It is required that a thoroughly cleaned blank hold a water film in the vertical position for approximately 15-20 seconds over the entire surface. Since the cleaned surface is easily oxidized, lamination is accomplished no later than 1 hour after surface preparation. Lamination The SPF-2 photoresist is laminated on an installation for application of dry film photoresist to both sides of a clean dry blank with the rollers being strictly para~'.lel and at a pressure of 0.5 kgf/cm. The temperature of the heating shoes should be maintained in the range of 115 + 5�C over the entire length of the shoes and should be monitored constantly to avoid polymerization of the SPF-2 in case of _ overheating. Poor adhesion of the SPF to the blank is observed if the temperature decreases to a value below that indicated. Nevertheless the nega.tive effect of temperature deviation from the given range can sometimes be compensated for by changing the rate of lamination. The latter is selected experimentally for each lot of SPF and is in the range of 0.7-1 m/min. The heating shoes and the laminating rollers should be completely clean, without photoresist residues on them, which are removed by methylchloroform. Methylene _ ~hloride must not be used for this since not the methylene chloride vapors affect the SPF-2 photoresist, but they also ~reak down the roller material. The shoes and rollers may not be cleaned mechanically since low-quality laminata.on which leads to retouching or rejection during subsequent processing is observed if they are damaged. Lamination is usually carried out on cold blanks. However, if the SPF-2 photore- sist is used for a"screen" or if it becomes subject to the effect of high temper- atures during subsequent production operations, lamination must be carried out on hot blanks (SO-100�C). To increase the adhesion ~f the photoresist to the blank, - 5 . FOR OFF[CIAL USE ONLY _ APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE OtdLY the latter is aged for not less than 30 minutes after lamination prior to the next operation. SPF-2 photoresist of type 2-60 or 2-40, laminated two times, is used to produce a photopolymer layer of great thickness while providing the "screen" process on blanks 1.5 mm thick with metallized hole diameter of 0.8-1.2 mm. The blank is aged for 30 minutes prior to the second lamination. The lavsan film which provides me~ - chanical protection and serves as a barrier against the effect of atmospheric oxy- gen on the photopolymer layer, is removed immediately prior to developing or during ' the second lamination. Exposure SPF-2 dry film photoresist is a negative material: it is polymerized on theexposed - section and it becomes insolubl2 in metal chloroform. SPF-2 photoresist has a thick photopolymer layer and is exposed through the protec- - tive lavsan film. Exposure is carried out in an exposure installation with colli- mated movable light source--flash lamps. Air-evacuating gaskets are used inside the vacuum frame for rapid removal of air and to provide good contact of a photo- template and the SPF. Heating in the vacuum frame is not permitted since it in- creases the rate of polymerization of the photoresist and leads to variation of the optimum exposure time, which is selected experimentally and is periodically checked: after replacement of the lamps, prior to the use of a new lot of SPF and after reglacement of the film in the vacuum frame. Determination of the aptimum exposure time is also important because the electrolytes of the galvanic bath are rapidly contaminated by organic impurities with short exposure and accordingly with insufficient polymerization of the exposed SPF. n A test-card phototemplate on which the width of the conductars is measured at var- ^ious points on its entire surface is used to determine the optimum exposure time. The blanks with the applied SPF-2 photoresist are exposed thorugh a test-plate negative with different speed of the flash lamp carriage. A blank with brilliant hard layer uf photoresist and with even unwashed edges of the conductors is selected after development. The measurements are made at the same points as on the template on these blanks by using a microscope. The exposure time at which conductors with clear edges and minimum deviation of their dimensions toward the positive side are produced is optimum. The time during which the conductor cross-sections on polished sections have plumb edges after exposure, development and galvanic buildup of inetal will be aptimum for the positive image method of printed-circuit card manufacture. Development Development of SPF-2 photoresist is based on dissolution and washout of the photo- resist layer not polymerized during exposure in methylchloroform. The developing installation provides rapid and complete removal of this layer under a stream of _ sprayed solvent. The developing time depends on the thickness of the light-sensi- tive layer, on the temperature of the developer which should be monitored and 6 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY should be in the range of 15-20�C and on the pressure at which the solvent is fed to the blank. The optimum developing time is selected experimentally and is taken as 1.5 times greater than the time required to remove tlie unexposed layer of photoresist. Z'he optimum developing time comprises ~0 seconds for SPF-2 photoresist of type 2-40 at pressure of approximately 2�105 Pa at the nozzles and at solvent temperature of 16-18�C. The photoresist is washed out during development of the positive image from deep channels; therefore, the developing time is increased by 5-15 ~econds depending on the density of the circuit pattern and the width of the conductors. The developing time, temperature of the developing solvent, the purity of the de- veloper and final flushing with cold water affect the quality of the developed image. Complete washing of the unexposed photoresist is nec:essary since even traces of it during subsequent operations lead to rejection. The contaminated solvent is partially broken down during purification (distilla- - tion) while the impurities and products of inethylchloroform decomposition affect the exposed photoresist and cause a reduction of the range of developing time. Therefore, no more than two parts of the distilled solvent and one part of puri- fied solvent must be used in the developing chamber, while only pure solvent must be used in the washing chamber. The solvent is contaminated not only by the diss~lved photoresist and impurities but also by water whose presence threatens low-quality developing. Rapid careful washing of the pattern by a stream of cold water under pressure~ is required after developing to remove residues of solvent before the developer be- gins to evaporate, after which the blank with developed image is protected by fine polishing powder with subsequent flushing with water. Becoming greasy--the unwet- tability of the copper surface on blank sections--occurs with low-quality develop- ing. A well-developed image should support a continuous film of water for 20-30 seconds. Introduction of regulators to the photopolymerizing composite which fa- cilitate the developing process is possible with further improvement of SPF-2 photoresist. Not only the chemical action of the solvent but also mechanical action, which is pr~vided by spraying methylene chloride under pressure at no less than 3�105 Pa in installations to remove the SPF are required for complete removal of the photo- resist from the exposed sections. But since methylene chloride has a negative effect on the dielectric properties of the base after etching the copper from the blank spaces of the card, the removal time should be minimum. Thus, the removal time for SPF-2 photoresist of type 2-40 at temperature of the removing solution of 16�C and at pressure of 3�105 Pa com- _ prises 25-30 seconds. The pattern is carefully flushed with water after removal of the photoresist with subsequent cleaning of the surface with polishing powder and flushing. The use of dry photoresists as a base to produce saturated conducting patterns on large printed-circuit cards makes it possible to solve the most i.mportant 7 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ON~,Y production problems in manufacture of modern microelectronic apparatus and computer equipment. Tt-lis conclusion has also been confirmed by the experience of leading foreign and domestic firms. Extensive development of these materials is a necessary condition to convert to - more pro:nising positive-image and combination methods of manufacturing two-sided and specifically of multi-layer printed-circuit cards. Dry photoresists find wide ' applic,~tion not only for subtractive but also for additive (semiadditive) - techniques. Preliminaiy investigations also show that the production processes using dry photo- resists a::e more easily subject to automation and find broad application in modern ASU TP [Automated production process control system] of printed-circuit cards. BIBLIOGRPAHY 1. Fedilova, A. A., Ye. P. Kotov and E. R. Yarvich, "Mnogosloynyye pechatnyye platy" [Multilayer printed-circuit cards], Moscow., Sovetskoye radio, 1977. 2. Kuznetsov, V. N., V. A. Pogost et al, "Some Chara~teristics of Usi;ig Dry Film Photoresists Developed by Organic Solvents and By '~Vater-Alkaline Solutions," VOPROSY RADIOELEKTRONIKI. TEKHNOIAGIYA PROIZVODSTVA 2 OBORUDOVANIYE, No 3, - 1978. - 3. Information Manual. Riston Photopolymer Resist, DuPont, 1972. COPYRIGHT: Izdatel'stvo "Mashinostroyeniye". "Pribory i s~stemy upravleniya", 19~31 G521. CS 0: 1863/155 8 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR l~Ft~i~^IAL USE: UNLY - MULTIMACHINE ANA MULTIPROCESSOR SYSTEMS Moscow NOVOYE V ZHIZNI, NAUKEy TF.RHNIKE: SERIYA "RADIOELEK~RONIKA I SVYAZ"': EVOLYUTSIYA VYCHISLITEL'NYKH SISTEM in Russian No 3, Mar 81 (signed to press 12 Feb 81) pp 44-45, 62 [Excerpts from chapter 6, "Multimachine and multiprocessor systems," and conclusion from the monthly "Radioelectronics and Communications" serial, "What's New in Life, Science and Technology": "Evolution of Computer Systems", by Yevgeniy Pavlovich Balashov, doctor of engineering science, professor, author of 12 books and 96 inven- tions, who specializes in computer technology; and Arkadiy Petrovich Chastikov, can- didate of engineering science, docent, author of over 40 scientific articles and many inventions, Izdatel'stvo "Znaniye", 38,260 copies, 64 pages] [Excerpts] Multimachine and Multiprocessor Systems In 1976, associates at the Institute of Mathematics, Siberian.Branch of the USSR Academy of Sciences together with the Severodonetsk Impul's Scientific Production - Association engineered the MINIMAKS project, a minimachine program-switched system. riINTMAKS is a homogeneous system with program-switched channels for communication - between the elementary machines (EM). Each elementary machine consists of a compu- ter complex (M-6000 or M-7000) to process data and a systems device for communica- tion between the elementary machines. It is interesting to note that the ratio of cost of the systems device and the cost of the MINIMAKS processor does not exceed 0.5, while this ratio for multiprocessor systems with a common bus with the RDR-11 computer processor is 1.2. The MINIMAKS system may function autonomously, as part of high-capacity concentrated computer complexes or as part of distributed computer networks. In the process, it may be successfully used both to solve scientific and economic prohlems, and to control grocesses. Among subsequent developments of homogeneous computer systems, we m~st mention the minimachine system (SUMMA) developed in 1977 ~nd based on the Elektronika 100 and Elektronika 100I domestic minicomputers, a system oriented to operation as part of automated systems for control of industrial processes. The Institute of Mathematics, Siberian Branch of the USSR Academy of Sciences, and the NITsEVT [Scientific Re- search Center for Electronic Computer Equipment) have now engineered a project for a general-purpose homogeneous computer system based on the YeS-1060 computer. Its overall speed is determined by the number of YeS-2060 processors and by ratings may be from 2 to 20 million operations per sec~:nd. 9 FOR OFFiCIAi. iJSF (1NT,Y APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFI~IAL USE ONLY The development and production in our country of homogeneous computer systems is a significant achievement in the field of building multimachine computer facilities with a programmable structure. Conclusion In the development of domestic computer technology, a great contribution was made by academicians S. A. Lehedev, V. M. Glushkov, A. A. Dorodnitsya, G. I. Marchuk, I. V. Prangishvili, V. S. Semenikhin, corresponding members of the USSR Academy of Sciences I. S. Bruk, V. S. Burtsev, Yu. A. Bazilevskiy, G. P. Lopato, A. A. Lyapunov, - A. P. Yershov, N. Ya. Matyukhin, B. N. Naumov, doctors of engineering science E. V. Yevreinov, M. A. Kartsev, A. M. Larionov, V. K. Levin, B. N. Malinovskiy, V. V. Przhiyalkovskiy, B. I. Rameyev, M. R. Shura-Bura and other researchers, designers and developpr.s of data processing facilities. _ Developers are now faced with a number of problems, among which the following may be mentioned: formation of new computer series, rational combination of hardware and software in computer systems, formation of advanced structures based on large-scale integrated circuits, further improvement of the design and technology base, etc. It should be noted that the socialist countries have found a successful path in the _ field of developing computer technology, development of a second series of YeS - ~unified system) computers with a throughput of millions of operations per second is nearing completion successfully, and the high-throughput computer systems E1'brus-1 - and E1'brus-2, M-10, YeS-1035 and Yes-1045 with matrix processors have been developed. COPYRIGHT: Izdatel'stvo "Znaniye", 1981 8545 CSO: 1863/162 10 FOR OFFICIAL USE Ot'dLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 1~1)R (N~I~I(7AL 11~N: ONI.Y BASIC C'riARACTERISTICS OF SOME DOMESTIC LSI MICROPROCESSOR COMPLEXES - - Moscow NOVOYE V ZHIZNT, NAUKE, TEKHNIKE: SERIYA "RADIOELEKTRONIKA I SVYAZ'": EVOLYUTSIYA VYCHISLITEL'NYKH SISTEM in Russian No 3, Mar 81 (signed to press 12 Feb 81) p 55 , [Tab]_e from chapter 7, "Evolution of Microprocessor Technology," from the monthly "Radioelectronics and Communications" serial, "What's New in Life, Science and Technology": "Evolution of Computer Systems", by Yevgeniy Pavlovich Balashov, doctor of engineering science, professor, author of 12 books and 96 inventions, who specializes in computer technology; and Arkadiy Petrovich Chastikov, candidate of engineering science, docent, author of over 40 scientific articles and many inven- tions, Izdatel'stvo "Znaniye", 38,260 copies, 64 pages) [Excerpt] Basic Characteristics of Some Domestic LSI Microprocessor Complexes Circuit T~pe Function Technology Word Speed, Number of Supply ~ize, t,~i s_ Instructions Voltage, V _ Bits (Micro- Instructions) K-536 (14 LSI) Various PMOS 8 10.0 149 -24; +1.5 K580IK80 Microprocessor NMOS 8 2.0 78 +S; +12; -5 K581IK1 Register ALU NMOS 16 1.6 +S; +12; -5 K581RU1 Microprogram ~OS 12 512 +5; +12; -5 Storage _ K582IK1 Processor I2L 4 1.5 1.2 Element K-583 (13 LSI) Various I2L 8 1.0 K5841K1 Microprocessor I2L 4 1.5 512 1.2 _ K-586 (3 LSI) Various NMOS 16 0.5 K587IK2 ALU CMOS 4 2.0 168 9 K587RP1 Control Memory CMOS 1.5 9 K587IK1 Data Exchange CMOS 8 1.5 60 9 K587IK3 Arithmetic ~OS 8 2.0 64 9 , Expander 564RU2A 256-bit Main QqOS 1 2.5 3-15 Memory 564RU2B 256-bit Main ~OS 1 3.5 3-15 Memory K530AP2 Bidirectional TTLDSh 4 0.03 Amplifier-Driver COPYRIGHT: Izdatel'stvo "Znaniye", 1981 8545 CSO: 1863/162 11 FnR OFFICiAI. I fSF. (1Ni.Y APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 hOR OFHICIAL USE ONLY P320 PROGRAMMABLE CALIBRATOR Moscow PF~LBORY I SISTEMY UPRAVLENIYA in Russian No 3, Mar 81 p 37 [Text] The P320 calibrator k~th manual and program control is designed for use in automated checking installations and also as an independent device for checking analog and digital devices operating on direct current. The device is a d~sk type, � portable and made from the designs of the ASET [expansion unknown]. - The calibrator provides for transmission of calibrated voltages and currents in the range from 1�10'S to 1.1�103 volts and from 10-9 to 1.1�10'1 A, respectively, and remote (program) control, including setting the maximum and the level of the output ~ parameter. The level of the output parameter is controlled in binary-digital codes 8- 4- 2-1. Maximtun Errors of Relative Value of Maximum Calibrated Voltaqes (Currents) Calibrated Voltages (Currents) 0.1 V 5�10'S Uk + 10 ukV 1 V 2�10-5 Uk + 10 ukV 10 V . 1� 10-5 Uk + 40 ~tkV 100 V 3�10-5 Uk + 500 ukV Up to 600 V 4�10-5 Uk + 5 mV 1,000 V ,~ove 600 V 5�10-5 Uk + 5 mV 1 ~ 2�10'S I~ + 0.01 ukA 10 mA 5�10'S Ik + 0.1 ukA 100 mA 5�10'5 Ik + 1 ukA Note. Uk and Ik are the established value of calibrated voltage and current. The maximum permissible main error of the relative value of calibrated voltages and - currents does not exceed the values indicated in the table. The main error is numbered for the temperature range of tn + 2�C, where tn is the temperature at which calibration was made. The output impedance of the calibrator is in the range of 10 V< 0.001 ohms. Th~ level of the variable coznponents in the 12 F4R OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400024412-2 . FOR OFFICiAL USE ONLY grequency band up to 100 kHz (a mean square value) at the calibrator output < 300 ~ (for range of 0.1-100 V) and < 1,000 uV (in the range of 1,000 V). The calibrator has an overload protection device and limiter of the input valtage (current) level. The power consumed by the device at nominal mains voltage < 150 V�A. The mains voltage is 220 + 22 V(at frequency of 50 + 0.5 Hz). The overall dimensions and mass are 488 X 535 X 250 mm and 26 kg. The device is operated at 10-35�C and at relative humidity up to 80 percent. - Calibrators are delivered by order to Soyuzglavpribor (117219, Moscow, V-219, ul. Krzhizhanovskiy, 16). . The incorrectness of the data prese.rited in the article of V. V. Korchin and V. T. Kuz'min NTTM-80, published in issu~ No 10 of our journal for 1980, is the operating reason for publication of the givea remarks. COPYRIGHT: Izdatel'stvo "Mashinostroyeniye". "Pribory i sistemy upravleniya", - 198.. - 6521 CSO: 1863/155 13 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-00850R040440020012-2 FOR OFFI~'IAL USE ONLY Jj NL4J COMPUTER HARD4lARE Moscow PRZBORY I SISTEMY UPRAVLENIYA in Russian No 1~ Ja~n 81 pp /~/-/+5 [Article by S.B. Abramov~ candidate of economic sciences: "New Hardware"] [Text] The USSR VDNKh (USSR Economic Achievementa Expoaition) Pavilion~ ~~Computer Equipment~~~ featuring theme expoaitions "Advanced Know~ow of the Kazan Computer � Plant in Perfecting Series Production of Computer Hardware~~ and ~~Mechanization and Automation of Computer Operationa" demonstrated new c~mputer hardware manufactured by enterprises of the Miniatry of Instrument Building~ Automation Fquipment~ and Control Systems. The article provides a brief description of the more interesting exhibits~ noted for the novel design-engineering solutions inc~rporated~ and the high level of technico-operational parameters displayed. - _ A desk-top terminal featuring aimplest information processing~ the ~~Iskra-900~~ is designed for the arithmetic processing of da.ta~ timekeeping~ automatic te~ephone s~,rltching-connecting~ simultaneous conversation among three members-au~scribera - of the telephone system~ and exchange of digi.tal information with the computer. Including the indicated f~Znctions i.n the terminal procludes the need for a group of single-function instruments~ (keyboard computexy cl.ock-calendar~ auto-dial telephone~ teletype)~ and reduces the number of operations involved in data - preparation for input into the computer. The terminal can be employed by upper and middle~-management links i.n institutions and enterprises from vaxious areas of the national economy~ and by operators in- volved in data preparation at remote peripheral points of the ASU and directly by production personnel. It ia uaed independently and in conjunetion with the M-6000 and SM computers. From a design standpoint~ the ~~Iskra-900~~ consists of a control console and inte~ face unit. The control console providea for the f~Znctions of t~he EKVM (keyboard computer)~ the clock-calendar~ auto-dial telephone~ and also, in conjunc- tion with the interface unit~ four terminal operational modes (EKVM~ auto-dial telephone~ clock-calendar~ terminal). Specifically, in the ~~EKVM~~ mode~ the terminal performs arithmetic operations~ chain-catenary operations~ and operations with conatants with allowance for the result c,haracter of the executed operation and overflo~r of the binary word format. - 14 FOR O~'FICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY _ In the ~~Terminal" mode~ the machine prov.Ldes for the trr~namiasion to the computer of data output from the keyboard~ selected from the operational memory~ EKVM computation results~ current time~ and also t~he receipt of data from the computer. Transmitted and ~eceived data are indicated on che terminal~s indicatos~ display. The electronic ticket counter outlet (EBKM)~ ~flskra-307"~ ~s designed for mechaniza- tion a.nd accounting of ticket-outlet operationa for passenger traffic on rail.road ~ tranaportation facilities. - integrated circuits (IS) with low level integration of the Ifl.55 _ series, the machine consista of a calculator~ tape perforation unit~ and a printer. Input of data is accomplished via an alphanumeric keyboard on the calculator and minicassettes~ with output to the II,T screen (during the procesa of ticket registra- tion)~ printer (travel document~ control tape)~ and tape perforator. - The "I skra-307~~ performs the following operations: accounting of monetary receipts in a overall summation register, and cashier tot~.l. receipts per shift~ an accounting _ of the number of travel documenta issued by category; an accounting of the cost and number of voided tickets; and the automatic price determination for vasious - types of travel ticketa. The machine "Iskra-3~7"~ in comparison earlier used machines of the same class of electromechanical machines has expanded :f~znctional capabilities: recording a large volume of data on a machine medium~ with ;ntbsequent compute~~ data processing~ lesaer dimensional size~ and lower weight. Tectuzica.l Characteristica of the "Iskra-307" Comput,erized Ticket Outlet (EBKM) " Number of counters: SUIIIIIl&T,}/ ~IDOI78ti8Ty~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~� 6 operational~~~~~~~~~~.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~� 1. Volume of counters in decimal places monetary~~~~~.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~� 6 - operational~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~� Volumes (capacities): OZU, bi~s .........................................o....l~K X 8 D10ZU~ bits ..............................................102/~ PZU, bits ..............................................8K X 16 r T, symbols .......................................e....256 - WLTAGE~ volts..... .......................................220 (50 Hz) The electronic control-registration (EKRM) nlskra-300~~ performs cash accounting operations in trade and public enterprise9. The machine incorporates laxge scale IS and a microprocessor. Data input is from an alphanumeric keyboard' output to a seven-digit built-in indicator and to a printing device (control check~ tape~ detachable check~ accounting register). Two documents axe simultaneously printed ou~--the check or statement~ and the control tape~ The ~~1 skra-300~~ performs the following operations: accounting~ individual total~ - 15 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY _ cnange~ multiplication, addition~ eubtraction~ repeat, and void. Use of the machine increases the productivity of accounting purchases with con- sumers~ as the cost of purchases is calculated by price~ weight~ or number. Technical Gharacteristics of the ~~Iskra-300~~ EKRM Informat,3.on volume~ OZU in bits .................................256 X 4 Printir~g speed~ lines per second ...................................2.5 CiOTlti1'OI. registers~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~e~~~~~} d1,~1.~'i Number of character positions pe.r line..............,~~~,~,~,~~~~~~16 Power lri jl8,tt8~ ri0't IDOr@ '~',~1gI1� . . . . . . . � � . . . � � � � � ..2~~ 1?imensions~ 1T1 1IlIll~~~~~~~~~~~~~~~~~~~s~~~~~~~~~~~~~~~~~~~~~~486X/.05X/.,.00 4leight~ in kilograms~ not more than ..............~.~~~~~~~~~,~~~~~~/~p Centralized control machine (MTsK)~ Series M-40-43~ is deaigned to solve control and management prot~lems for processes in various i.nduatrial sectors and constitutea the latest development of Series M-~.0 machines~ which are based upon microprogrammed control. The Series M-/~0-1~3 machine is employed in enterpriaes ~rith continuous end diacrete type production (enterprises with ASU controlled machinery~ control of inetal quality for hot and cold rolling~ control and mar~agement of individual functional groups of power units~ small and medium substations~ and others). The machine is prcduced in four versions (see table). The first two versions axe u sed by enterprises with continuous and discrete production~ the latter two are used only by enterprisea with distinct type production. The basic set of machines includes a central control unit (WTs)~ also an input- output unit~ technological data print unit~ input-output of diacrete data unit~ telemechani cal communications (UTS)~ external memory (UVP-43)~ alphanumerical terminal, and an indication conaole. riodification Number of Maximum number of Capacity of high level module pointa of external ZU. Kbits signals discrete i_ nput out~ut M-4~-43/2 12g 352/z2 160/16 16 r~1-4o-43/z1 M-4~-43 3 102 64 320 32 16 i~~-4~-43/31 _ By design veraion~ the MTsK of the N~40-l+3 series belongs to the ASUT-M syatem. Communication With external devices ia accomplished by a 2K communication rank. 16 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R000400024412-2 FUR OFF[CIAL USE ONLY Tec,hnical Characteristics of the MTsK M-40-43 Maximum number of external units (peripherals) connected to: WTs............... ...........................................up to 16 WTs via TJTS ..................................................up to 28 Maximum number: - High level analog input signals ...............................up to 512 Discrete input data points....���.� ...........................up to 1/+00 ~9aximum s~eed for execution of microcommands in thousands of operations per second...........o ................................5G0 Capacity in bitea ~8T1C~OD1 acceas file ~U~ .~i.3/~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~Z.bK N:emory uni.t: Glxrrent ..........................................................2K Permanent~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~s~~~~~~~$K Voltage~ volts.................. ................................320/220 (50 Hz) Required power~ in V.A.(voltr-amperes) ...............................3500 I1a11 operating time in hours~ TlOt'i 1.88$ ti~l&21~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ Dimensions in mm ...........................................650X6787Q 600 Weight~ kilograms ...................................................2000 The SPK-80 aperture punchcard search selector is designed for the rapid aearch of data input in the form of perforations on atandard 80-column punchcards~ and ia employed in searches for patents and inventors~ certificates~ information in technical. information officea~ and for design analogs for the use of reac~y solutions or the study of e~dsting teclmical levels of tasks solved~ i.n personnel departments and other institutions. The card feed and photoelectric method of reading of data provides for a signifi- cant reduction in the wear of punchcards~ an extension of their service life~ and the single-ghase feeding net~ low weight~ and sma71 dimensions and low noise level permit the use of the selector in any organic equipment complex~ ASU or autonomously in any sub-element where it is necessary to aelect from a deck of cards with the required data. The selector selects punchcards with total coincidence of information on the selected punchcard w,~~h information input to memory~ the coincidence of iriforma- tion of one of the zones of the deaignated punc,hcard with i.nformation of the zone input to memory~ and ~:~e coincidence of information from a given column of the selected punchcard ~:ven if only one of the characters input to memory for that seme Golumn. Technical Characteristics of the SPK-SO Tec,hnical speed 121 CSTC~S/II11T1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~j.~0 Capacity of magazines in cards: receiving~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~W~ feeding .........................................................~.OQ Number of simultaneously read positiona ................~,.,,~~~,~~~~12 Voltage~ i.n volts ..................................................220 (50 Hz) Required power~ in V.A .............................................500 Dimensions 121 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~7v0X5W~j,W Weight in kg .......................................................100 COPYRIGHT: Izda,tel'stvo ~~Mashinostroyeniye". "Priboi-y i sistemy upravleniya"~ 1981 8851 CSO: 1863/142 17 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAG USE ONLY SOFTWARE UDC 681.328 SYSTEM FOR COMPLEX DEBUGGING OF ELEKTRONIKA 55-12 MICROCOMPUTER AASED ON FUNCTIONAL MICROPRUCESSOR MODULES Moscow PRIBORY I SISTEMY UPRAVLENIYA in Russian No 3, Mar 81 pp 36-37 [Article by Candidate of Technical Sciences V. V. Sumin and engineers V. N. Syrtsev, A. F3. Vasil'yev and N. M. Kalinin] [Text] Single-board microcomputers of the Elektronika S5 family are now a promis- ing technical base for construction of various automated systems. It is very im- portant in this case for the user to have the capability of cor.:plex debugging of the equipment and which is more timely the program parts of the system under development. One of the variants of constructing a debugging system for the Elektronika S5-12 microcomputer on the basis of functional microprocessor modules serially produced by our industry is described in the article. The complex debugging system is de- signed for preparation, editing and debugging of programs for systems constructed on the basis of the Elektronika S5-12 microcomputer. The system permits entry of information from a type FS 1501 photoreading device, program editing by means of keyboard and console machine of the Consul-260 type, derivation of information to the PL-20 printer and light diode displays and listing o� the programs being de- bugged can be achieved if necessary. The debugging system is capable of successfully replacing the industrial prototype system constructed, for example, on the basis of the multiboard Elektronika SS-02 microcomputer (for a specific class of problems). A block diagram of the debugging system OS is presented in Figure 1. The system includes a microcomputer of the Elektronika SS-12 type, functional microprocessor modules OZU [Internal storage] Elektronika 55-125 with capacity of 2,048 X 16 bits and TsW [Digital input-output] Elektronika 55-122 (it has four each digital in- - puts and outputs), read-only memory PZU with capacity of 256 X 16 bits, matching devices SU, light-emitting diode displays SID, peripheral devices (electronic con- sole machine EPM of the Consul-260 type, perforator P of the PL-20 type and photo- reading device FU of the FS 1501 type) and a keyboard K1. Without describing in detail the organization and information exchange in the sys- tem, let us note only some of its features. Thus, the use of the Elektronika 55-122 TsW for exchange of the debugging system with the periphery permits 18 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R000400024412-2 FOR OFFICIAL USE ONLY ~ l I ~ I ~y ' I _ I ~ ' ~ I ~1~ 1. ~ ~y ~ MuRpr o ~ - ~ p3y -3BM ~ I I ~ ~ ~ ~ . I I I ~y I I I ~ ' ues ~ i ~ i ~ C9 nepurpepuu~2~ ~ I I L-- - - - - 1 3lIM 11 ~SI Ifn CHQ Figure 1. Block Diagram of Debugging System: ShA, ShI and Upr-- address, information and control buses Key: 1. Microcomputer 2. Peripheral switching devices preservation of the digital inputs-outputs by the microcomputer itself for working with the automation facility. The use of the Elektronika 55-125 OZU module with capacity of 2K (or 4K) permits storage of user working programs and the program for exchange with the console machine. The read-only memory PZU with capacity of 256 words is realized on a micro-OZU of type 561RU2 with buffer power supply. It con- sumes less than 20 ~.~A in the storage mode; the information is retained for 5,000 hours after cutoff of the power supply. The booster storage battery is precharged automatically when the system power supply is switched on. A dispatcher program - which services the work of the system with keyboard K1, SID display block, perfor- ator P and FS 1501 transmitter is written into the PZU device. The keyboard of the - debu~3ging system consists of keys of the operating code ("Writing of initial ad- dress," "Writing by address," "Reading by address" and "Function") and of a key- board of a 16-digit symbol field. The "Function" operation is introduced to expand the capabilities of the system without additional apparatus expenditures. Check addition, file shift, initial start of program from given address, entry from the - photoreading device, printout of information on punch tape and conversion to pro- gram of working with Consul-250 machine can be accnmplished by using this opera- tion. 5pecific functions are introduced by means of the keys of the 16-digit key- board field. When realizing the complex debugging system, the greatest difficulties may be en- countered in organizing the integration of the Elektronika 55-122 TsW module with the peripheral devices. One of the versions of this organization is presented in Figures 2. The Elektronika SS-12 microcomputer continuously interrogates the keyboard of the complex (two keyboard racks are related to a 16-digit code and one is related to 19 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R000400024412-2 FOR OFFICIAL USE ONLY r ~ ~ z t ~ e . ,9 J 4~� ~ ' ' � 9.~nbA 8 d ' d � .f ,~ynKyue~2~ 1)H r~ N ~ ~1 11 Ij N e' ~ g , ~ C RC 2 � y~ 3 O ~ CYI ~Cun 1-d p ~p, N 4 ' CYB unm E�u pq~p K f5 lsol e 3 � vc tu i. uM ~ ~ Cmon' S Z,CmaOm� ~C R r ~na m _ ~ ~n00~~P1 +Unun ~ /!CT o un � M p nq R! p u~: d-& o.t ' M�20 4 P2 ~Z 3/!M t~ _ ; i ~ ~ l� pa~p ( 6 ) ' ~ ~ i ~ ~ P9 ~ i ~ 7 ~ +u I(9 ~a 3pM ynpal,l ` tf, C num �u pa~p a~exmpowaa- ~ a 3nN ryumqM4 37?M ~ pa~p ~ i ~ ~ 3/IM unp /6~dpoJp ii JI7Nun~l�dpa7p ~8~ ~ ~n �u r Nnq~opNauu~ ~ ~ JqN un~ om J/7M - Figure 2. Diagram of Integration of TsW Module and Periphery: Tsvykhl- Tsvykh4--digital outputs 1-4 of TsW module; Tsvkhl-Tsvkh3-- digital inputs 1-3 of TsW module; ZP, Cht, ZpBA and Function-- operating codes "Writing by address," "Reading by address," "Writing of initial address" and "Function": a, b, g-- light-emitting diode display buses; SU1-SU8--matching devices in photoreader circuit; R1-R9--encapsulated solid-state switches of control of perfora~or and selection of decoder column of EPM machine; K1-K9--encapsulated solid-state switch contacts; K11- K116--case for selecting the decoder column of EPM machine; RG--recorder-distributor; T--f lip-flop for control of FS 1501 - transmitter operation; EPM inf lr-EPM inf 7r--information from EPM encoder; FS--photoreading device; P--perforator Key: 1. Elektronika SS-122 5. Strobe 2. Function 6. Digit 3. Start 7. Control of EPM electromagnets 4. Stop 8. Information from EPM the operating code) through the diqital output Tsvykhl of the Elektronika S5-122 module. If not a single key is pressed at a given moment, the microcomputer accom- plishes dynamic display of the contents of the display buffer in 16-digit code through digital output Tsvykh2. Digital output Tsvykh3 of the TsVV module is used to connect the perforator to the debugging system. The form of retrieving the information is determined by the sub- program for working with the perforator. 20 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY Information from the FS1501 transmitter through the matching devices is entered through digital input Tsvkh2 of the module in parallel code; tr.e form of the enter- ed inforutation is determined by the subprogram. Th2 status of the console machine is interrogated through digital input Tsvkh3. The code for control of the corresponding electromagnets of the printing mechanism = is transmitted through diqital outputs Tsvykh3 and Tsvykh4 of the module as a function of the type of control code. If a code comes in that corresponds to a 16-digit symbol, the microcomputer prints this symbol and enters it in the infor- mation buffer after conversion. The described debugging system permits one to realize two important debugging func- tions in development of automated systems based on single-board microcomputers of the Elektronika S5 family which are determined by the location of the microcomputer. If the latter is located in the system to be debugged (a microcomputer is then un- necessary in the debugging complex), the periphery of the debugging complex works with the user program and permits complete debugging. If trie microcomputer is contained in the debugging complex (the system to be debugged may then have no microcomputer), the periphery of the complex is connected by the microcomputer to the facility equipment and permits debugging of the relationship of the facility and microcomputer. _ Finally, let us note two additional circumstances related to the described system. First, the use of serially produced Elektronika 55-122 and Elektronika S5-125 func- tional microprocessor modules, hardware- and software-compatible with the Elek- tronika 55-12 microcomputer, permits significant facilitation in realization of the debugging system and completion of it without some significant circuit developments. Second, although the described system assumes the presence of a rather developed _ periphery, debugging can also be carried out in minimum peripheral configuration, for example, by using only a keyboard and light-emitting diode displays. The described system replaces the scarce prototype systems based on multiboard models of Elektronika 55-02 type for most practical purposes. COPYRIGHT: Izdatel'stvo "Mashinostroyeniye". "Pribory i sistemy upravleniya", 1981 6521 _ CS 0: 1863/155 21 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-00850R040440020012-2 ! FOR OFFICIAL USE ONLY - UDC 658.012.011.56: MAIN TRENDS IN FORMATION OF AN ALGORITI~M AND PROGRAM FUND FOR OASU Moscow PRIBORY I SISTEMY UPRAVLENIYA in Russian No 3, Mar 81 p 5 [Article by engineer S. M. Khovanskaya] _ [Text] An increase of the efficiency of introduced OASU [Automated sector control system] and their effect on the improvement of the technical and economic operating indicators of ministry and agency subdivis ions can be achieved by standardizaticn of OASU components and automation of its design. The problem of standardization includes formation of a system which satisfies the specifics of a given facility from standard components. The use of inethods of standardization in design of OASU permits a reduction in the cost of developments, a reduction of the periods of systems development and a reduction in the number of OASU developers. Standardization of the control system components in our country is being developed in three directions: development of standard ASU [Automated control system], uni- fication and standardization of individual parts of functional and support subsys- tems and construction of ASU by use of applied program packs (PPP). Practice shows that development of standard ASU creates large complexities and is economically unfeasible since there are no completely identical ministries in na- ture. Consequently, two methods remain: unification and standardization of the individual parts of an OASU and design of OASU usinq PPP. This problem can be solved through extensive use of a centralized algorithm and program fund (TsFAP). - The PPP now available in TsFAP and oriented toward use in OASU can be divided into four groups by designation. 1. Means of organization and management of the information base of the OASU. 2. Functional subsystems. 3. Means of programming automation, 4. Means of organizing the calculating process. The "Glossary organization and managment" PPP, the "Driada" data bank, the "Ac- _ counting" PPP and the "Data" PPP belong to the first group. - 22 FOR OFFICIAL iJSE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 - FOR OFE'iCIAL USE ONLY OASU subsystems developed within the ASU-pribor system are related to functional subsystems. These subsystems can be utilized to the maximum in design of OASU of machine building. Analysis of the possibility of using subsystems in other sector _ groups (the light, local and mining industries) showed that it is impossible to in~roduce programs availabie in the fund without modifications. The circ.umstanoA that conversion to the two- and three-section structure of management has still not been completed in the ministries until now also creates great difficulty in use of the programs . The "Compiler" PPP and the "Monitor" PPP ar.e included in the group of PPP that automate the process of program development and debugging. An importan~ problem is development of an effective technology of design with ex- tensive use of the algorithm and program fund. One of the forms of automation of OASU design may be development of a so-called systems fund by groups of sectors which, unlike existing funds, should be a unified organizational system. The following facilities may be included in the systems fund: general-purpose PPP to which PPP that realize development and management of glossaries and files of ~ normative and reference information are related, development and management of data bases of current and archival information and printout of output forms and display of information; functional PPP that include those which enter and monitor account- ing information as it comes in with regard to functional features and calculation of the output indicators of PPP for orqanization and management of the c alculating - process; and a system for teaching how to work with PPP. - The developers of OASU using the systems fund will be able to realize specific management functions. In this case the degree of encompassing OASU prob lem~s with facilities of the systems fund may be close to 100 percent. Therefore, formation of an algorithm and program fund for OASU should be planned _ and purposeful. There should be no PPP in the fund that dc~ not have a high degree of universality with respect to specific functional solutions and that do not have means of adaptation to a specific medium of use by adjustment for given parameters. The need to meet the named requirements generates yet another problem: development of specific tests that check not only the functional but also the operational capa- bilities of PPP. Investigations are now being conducted by a number of scientific research insti- tutes (VNIPI [All-Union Scientific Research and Planning Institute] of OASU, Moscow, VNIIugol' [All-Union Scientific Research Institute of Coal] , Moscow, NPO [Scientific Production Association] Pishchepromavtomatika, Odessa, and so on) to develop standard software for OASU for its own sector groups,.which will undoubted- ly serve as the basis for organization of systems funds of OASU. COPYRIGHT: Izdatel'stvo "Mashinos~royeniye". "Pribory i sistemy upravleniya", 1981 6521 CSO: 1863/155 23 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY - unc ~19.6 COMPARATIVE COURSE OF THE PL~1 I,ANGUAGE (ON THE BASIS OF ALGOL-60~ Moscow SRAVNITEL'NYY KURS YAZYKA PI,~1 (NA OSNOVE AI~OLA-60) in Russian 1980 (signed to press 17 Apr 80~ pp 6-8 [Foreword from book "Comparative Course of the PL~1 Langua,ge (On the Basls of ALGOL-60)" by Yuriy Mikha.ylovich Bezborodov, in the series "Bibliotechka Program- mista", Main Editorial Boa.rd of Physical and Ma,thema,tica,l Literature ~ Izdatel' - stvo "Nauka", 65,000 copies, 192 pa,~es] [Text] In the last 15 years the ma,in, and at the begiruzing even the only algorithmic language in the Soviet Union was ALGOL-60, used both for teaching the principles of programming and for solving computational problems on computers. In recent years, however~ after the appearance of YeS computers, PL~1 (Programming Language) , which also is an algorithiai.c language, has become more and more widely distributed. In contrast with AIGOL-60, which is intended for the solution of problems of numerical analysis, PL~1 is a universa,l langua,ge which permits, in ad- dition, formulating solutions of commercial problems, problems in processing sym- bolic and binary information and ASU problems. It can be carisidered, in particu- lar, tha.t PL~1 combines the possibilities of such widespread la.nguages as ALGOL, FORTRAN and COBOL. On YeS ma,chines, PL~1 replaces ALG(7L in pra,ctica.l work, dis- placing it, which can be explaine~ not only by thie universality of PL~1 and its other merits, but also by major inad.equacies of the translator from ALGOL-60 in the YeS computer (ineffectiveness of the `transla,tor, inconveniences of the input language and especially of the input~output operators~ the impossibility of cauplin~ with other a].gorithmic languages and the absence of packages of applied programs). The present course of the PL~1 langua,ge is intended to ease for programmers the transition in their work from other computers to YeS computers and is prima.rily designed for programmers who ha,ve experience in programming in ALGOL-60 and want to become acqua.inted in a short time with the possibilities of PL~1 and to learn to program in tha,t language for YeS computer operating systems. In the course an attempt ha,s been made in teaching programming in PL~1 to use in a very essential manner the skill and experience in programming in AIGOL which the read- ers alrea,d.y have. True, besides practica,l experience, acquaintance with the terminology and syntax of ALGOL-60 also are required of the reader in tha,t ca,se [1, 2]. In Chapter 1 of the course the concepts of ALGOI,-60 and PL~1 are compared in such a way as to give the reader a conception about the creation axzd differences in the 24 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY designs of ALGOL-60 (in the standaxd versi.on) and PL~1 (for YeS computer operating systems) which have the same ~znctional possibilities. The used method of compar- isan wiil ena,ble the reader to ra,pidly sta,rt compiling programs in PL~1 within the _ range of its linguistic possibilities, whicY, intersects' the possibilities of AIGOL~ At the end of Cha,pter 1 information is presented which will help the reader prepare his own program for translation and rea.doff on the ma,chine. In Cha,pter 2 the PL~1 concepts introduced in Cha,pter 1 axe refined and, in a,ddi- tion, new concepts are introduced which are as it were a logica,l development of the ~ concepts of AI~OI,. Acquaintance with this cha,pter will enable the rea.der to use PL~1 means similar to ALGOL means in programming and will also help him to apply in complex concepts of PL~l introduced in Cha.pter 1. In Cha,pter 3 PL~l concepts are stated which have no direct analogues with ALGOL concepts. Without them the picture of the main PL~1 resources would be incomplete and it would not be possible to compile effective programs in a broad enough area of application. At the end of the chapter those PL~l possibil~ties are listed which have not been dealt with in this course and with which the reader ca.n become acquainted from the available literature [3~4,5,10] or for very complex concepts, - from documentation for YeS computer operating systems ~6,7~8,9]� In Cha,pter 4~, a reference cha,pter, a summary is iven of built-in PL~1 functions and conversions, tables of PL~1 syntactic forms ~one of them in comparison with ALGOL)~ a list of symbols used and also other reference ma,terials. All the chapters (except the last) are accompanied by exercises, the answers to which are presented at the end of the book (sometimes one of the possible answers is given). The solutions of exercises and analysis of the examples given in the presentation of the ma.terials play a very important role in the study of the resources of the and the acquisition of skills in compiling programs in PL~1. - Along with examples and exercises~ another means of checking the correctness and completeness of understanding of the studied concepts of the language ca,n be reducible syntactic forms, which express some properties of those concepts more clearly and graphica.lly tha.n when verbal descriptions axe used. Ana,lysis of, and, later, reference to those forms ought to prevent the appearance of syntactic errors in programs. Let us note once more that the course is oriented mainly toward practical program- mers who need not so much to become acquainted with peculiarities of the language or to study all its concepts~ as to acquire fairly ra.pidly the ability to compile real programs in it. Taking into consideration also that it is assumed tha.t the reader ha.s some experience in programming in ALGOZ, a language in a certain respect very similar to PL~l, the author thinks that a certain brevity in the account (com- pensa,ted by examples, forms and exercises~ will be desirable~for such readers. In the present course the PL~l langua,ge is described on the level of a specific representation as the input language of a version F translator for YeS computer operating systems. 25 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY The author is obliged to all who contributed to some degree to the writing and publication of th3.s book, The author especially thanks N. P. Trifonov for detailed and constructive criticism of the manuscript, and also T. N. Moshonkiva and A. K h.-M, Aldakov for numerous useflzl discussions and constant help in the work. COPYRIGHT: Izdatel'stvo "Nauka". Glavnaya red.aktsiya fiziko-matema,ticheskoy literatury, 1980 2174 cso: 1863/156 26 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY UDC 681.3.06s678 USING DATA BASES IN THE ASU TP OF LARGE-CAPACITY UNITS Moscow PRIBORY I SISTEMY UPRAVLENIYA in Russian No 3, Mar 81 pp 1-3 [Article by engineer s V. K, Sheremet'yev and A. I. Vikhter, USSR, and H. Berger, GDR] . ~ ~ [TextJ Separation of Proqrams and Data as a Basis for Standardization The applied program packsof ASU TP [Automated Production Process Control System] are rather stable in functional composition. Thus, for example, the interrogation, processing and information display modules are a realization of the simplest in- formation processing formulas and are practically identical by content in many systems. Nevertheless a la.rge number of diverse systems are being developed that perform similar functions determined to a significant degree by the different struc- ture of the information tiles. In most cases data organization is natural and is the result of pro grammers' use of individual program-writing procedures. Only the data to be processed change when a specific class of ASU TP is introduced at domestic facilities. Therefore, allocation of the variable part to a special domain, standardization of the data structure and development of facilities for formation of the data base (BD) for each specific facility should be the tasks of the systems programmer responsible for the overall ideology of software (PO) construction. Conversion of a sys tem to a domestic facility should not require significant ~ changes of software. Determination of the processes for development of programs from formation of the data base results in development of the mechanism for combin- ing the programs and datain real time (RMV). To do this, methods of constructing the data structures, organization of access to data at logic and physical levels used in general-purpose information systems, the basis of which is the concept of the data bank [1, 2], can be used in development of ASU TP software. Specifically, the language of processing control to be considered is close to the languages of data description, while the language of operator-data base communication is similar to the languages of data manipulation utilized in data banks. _ Specialists of the USSR and the GDR developed a system using a data base on the basis of the M-6000 computer complex within the framework of the intergovernmental agreement between these cowntries to develop installations of the Polimir type [3, 4J. ~ 27 - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY The characteristics of using a data base on the example of the ASU TP of the Polimir installation for polyethylene production are considered in the article. � Special attention is devoted to design of the data base structure, organization of _ access to it at logic and physical levels and formation of the data base using special language facilities. ~tructure of the Data Base The structure of the data base and organization of access to it are shown in the . figure. The data base consists of a core, tables for describing its structure that provide access to the core at the logic level and access lists SP1-SPn to the core at the physical level. .-~l~ ~(2) ' Qocmyn Ho ~otuvrt,roN y~o6Ht E PB Aalmyn Nc muiuvecraM ~pOJHI OntPO/llop ~3~ r-_-------~ ~ nD, - - ~ np ~ __J ~J rr-== ----P~ _ �P_e--- .~voAeildl (4) ~ i rann ~ Pr r~~ c~� ~ ~pynnoQei~ ~1 --------L- ~ .~o~poKmtputmdA'u I ~ ~ r- - ~ doarlyno,rl~Q TafO ~ i ~5~ I 1 ~y ~ ~ ~ Po~nvd 2~ ~ I TaTp ~ ~ ik,o ns�~n`d ~ uNduSudyo~e~IiJ[, ~ 6ror .rapa,rmeputmu,rQ s.o.:~~ i nopaMtmpo6 ~ I da~myno I 01 16~0"~ I I ~ I HO,IOIUVICKONI I nm i I ~ ~ ll~osyt ~81~ ~ Adpo 6Q 9 I I . I ~ 9voQeNe J~ 1 Q) I HOK I ~ ~OynnoBeie xadd,~. I . ~ I mrpucmurd u~'pd L-------- JJ~O/A,rUnopOMfmp00' Structure of Data Base and Organization of Access to It: RG--generation mode; RRV--real-time mode Key: _ 1, Real-t+*�Q ~^^n~~ at logic level 2. Access at physical level 3. Operator 4. Level 1: group characteristics of access to data base 5. Certificate heading 6. Block 7. Level 2: individual characteristics of parameters 8. Tables of access at logic level 9. Data base core 10. Level 3: group characteristics of parameter processing The data base core is a combination of the certificates of the technological param- eters and the common constant files. The certificate of the technological param- eter contains all data or references to the data which are needed to process the parameter. The common constant file (MOK) combines the constants repeated for in- dividual certificates. The tables for describing the core structure identify the individual structural components of the data base. 28 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY The information in the data base is distributed in three levels (see the figure). The table of the parameter lists (TaPP) and lists SP1-SPn form the lists of iden- tifiers and the relative addresses of the certificates of the technological param- eters and are the group characteristics of user access to the data base. The set of certificates P1-Pm of the technological parameters in the data base core and the tables of access to individual elements of the certificate characterize the individual features of processing indivudal parameters. All the repeated charac- teristics for groups of parameters are allocated to the lower level in the MOK. Organization of the data base permits collective access of user programs PR1-PRn and of the operator to the i.nformation concentrated in the data base core. Structure of the Certificate Each certificate consists of a title and several units, the number of which is equal to the number of types of processing required for a given parameter. The title of the certificate serves as the external interface required to organize ac- cess of information users to the data base. It also contains information on the state of the measuring circuits of sensors of the corresponding parameters. Each _ block contained in the corresponding certificate is the internal interface with respect to the standard primary or secondary processing operator. The certificate " block actually comprises part of the standard operator and contains coefticients and placings and also the results of the work of a standard operator. These data also provide external access for making changes and to print out the results. Moreover, the block title contains features for control of the standard operator work. On the whole, the certificate is a set of external and internal program in- terfaces which link the operation of the software system. The modular principle of constructing the certificate provides independence of separate standard processing operators from the overall data structure and the specific use of the sys~em. The structure of the data base core guarantees siznple access to the information and correction of it. The information available in the certificate can arbitrarily be s~parated into control and that to be processed. Control information is used to organize preliminary processing of the information, while information to be proc- essed contains the input data and the results of processing. Moreover, one can distinguish permanent, conditional-permanent and variable information components. Permanent and conditional-permanent parts are formed during initial generation of the data base. Conditional-permanent information can be changed in the RMV by using the access equipment. Variable information reflects the status of processing the information coming from the sensors. Access to the Data Base A description of the core structure is stored in the data base to provide independ- ence of user programs from its structure and for convenience of access to the data base. The data in the base core have structures of several types: information component, certificate block consisting of the components and the certificate which includes the title and a number of blocks. 29 FOR OFFICIAL USE ONLY _ APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY Description of the core structure (tables of access at the logic level) is a com- plex of retrieval tables (TaPP, TaTO and TaTD) to which the data required to re- trieve in~ormation in the data base core are reduced. The table of the lists of parameters contains identifiers of the technological parameters and data for retrieval of the certificates corresponding to them, the table of the types of operators (TaTO) contai.ns identifiers of the ty,pes of stan- dard processing operators and the corresponding certificate blocks and the table of types of data (TaTD) contains identifiers of the types of data that are the ele- mentary units of information from which the sub-blocks and titles of the certif- icates are formed. Each of the tables describes the types of str.uctures of a specific level and consists of blocks of identical structure fixed in value and _ which form in combination the table characterizing the entire set of types. Access to the data base core is organized in a different manner as a function of the information user. To increase speed, the access of user programs to the data _ base is gained at the physical level using lists SP1-SPn of the relative addresses of the required cells. Thus, retrieval of information in real time is essentially excluded. Lists of relative addresses are formed by the generation system using the tables which describe the structure of the data base core. When service personnel responsible for modification of the data processing system with regard to changing a production situation gain access tr~ the data base, data retrieval occurs in real time according to the tables for description of structure. This permits one to use parameter identifiers. The rate of access should have no special significance since this access does not have high priority. Thus, the rate of access to the data is provided in the first case and convenience of access is provided in the second case, although the process of linking programs to data itself is identical. The difference is only in the fact that information retrieval is related to the generation process in the first case. The distinguishing feature of the data base for an ASU TP is joining of the refer- ence function and the information restoration function. To economize on machine resources, the information coming from the sensors is restored directly prior to being given to the user. Thus, interrogation and preliminary processing of in- formation are aooomplished only if there is a corresponding request. Processing Assignment Language A special language similar in form to forms of the "fill in the blank" type, is used to enter input data and to assign of it. But unlike forms with fixed positions and riqid format of macroinstructions, it is similar to free for- - mat languages. The immediate significance of the data follows immediately after the identifier which determines the type of input data. A restriction is observ- ance of the required sequence of identifiers. Precise observance of the format aiid the number of spaces between positions is not obligatory. The processing assignment for each production parameter is described alternately in this language. The data required to generate the tables of access to the data base and the titles of the certificates are described in the first part of the 30 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFYCIAL USE ONLY assignment and then the data required to realize individual types of process- ing (correction of flow rates, scal,ing and so on) are described in specific se- quence. Up to 25 types of processing of parameters is provided. Each parameter utilizes its own set of types of processing. Thus, the sequence of the types of processing determines the assignment for processing each parameter. Data Base Generation Information entered by means of the processing assignment language is converted during generation and the data base core and some tables of access to it are formed. Thus static adjustment of the system for specific parameters of the facility is accomplished. The generation system has a modular structure and consists of a core and set of generators, each of which generates blocks of specific types. The set of generated blocks can be changed by replacing the generators. Generation of an entire system is accomplished in five steps. The first two steps are used to convert the expressions of the input language to some intermediate language in which the specific structure of the data base is not yet reflected. Conversion from the intermediate language to the internal structure of the data base is completed only in the fourth step of generation. Z'herefore, utilizing dif- ferent modifications of fourth-step programs, one can create different versions of data base structure. The software system is qenerated by using both the pack and dialogue modes. The latter is used to control the generation modes and operational corrections upon detection of errors. The system can be generated directly at the computer complex installed at a facil- ity to permit one to make chanqes directly during introduction ~f the system at the production facility. - Language of Maintenance Personnel Requests A special language is used to correct the data base and to organize access of main- tenance personnel to it. The language facilities permit callup onto the display screen and to print information from the data base that characterizes the status of individual units and of the installation as a whole in digital and graphical form, to call up complete information from the data base about a single production param- eter, including the preliminary processing constants, results of processing and so on, check and modification of individual components of the data base, inclusion and - exclusion of any production parameter to and from processing, for example, with regard to repair of a sensor, receipt of a list of the measurement points related to a single cycle and transfer of a measuring point from one interrogation cycle to another, elimination or inclusion of individual functions of the system, pro- � grams for display o utput or for printing individual forms of information display, printout of information about malfunctioning sensors upon callup and entry and changing of the time parameters of the system (entry of systems time, changing the time of information restoration on the display and so on). ~ _ The language is open for increasing the number of requests. The main components of the language are the type of request, the identifiers of the certificates of 31 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY production parameters, the blocks of the certificate and the components of the blocks. When individual components of the data base are summoned and corrected, tables of access at the logic level are used (see figure). Z'hus, for example, - the identifier T 6036 GA GI is used to retrieve the upper temperature setting with number 6036. The certificate is retrieved by using the TaPP according to the designation of production parameter T 6036, the required block is retrieved using the TaTO and the name of the GA block (the block of exceeding emergency boundaries) and the component inside the block is retrieved by using the TaTD and the name of component GI (the upper emergency boundary). In thvse cases when high speed is required, for example, when restoring information on the screen, access is gained at the physical level. Conclusions A system in which a data base was used as the basis was checked in the confection- ery shop of the Production Association Polimir imeni 50-letiya Belorusskoy SSR - (Novopolotsk). The following functions were realized in the system: interrogation of the low- and medium-level sensors, preliminary processing of analog information, diagnosis of the operation of the inteirogation and preliminary processing subsys- tem, recording of malfunctions in the circuit of the monitoring and measuring de- vices, signalling of deviations from the norm, writing a report of the course of the F,.^oduction process, presenting information to the operator on displays and calculation of secondary indicators. The use of a data base provided convenience of static and dynamic adjustment of _ ~he system, opera~ional rearrangement of the latter as changes were made in the instrument ancl ~r;~duction equipment and a considerable saving of inemory. Static adjustment of ~i~kze data base for processing specific production parameters and for realization ~f a~iven set of functions for each parameter are accomplished by using a generation ~~.~stem. Dynamic rearrangement of the operation of the system in real - time is acc~::1_.~.~.ished by entry of tasks through the display. Tests confirmed the correctness of the adopted engineering solutions. The system - as a whole showed good adaptability to the conditions of the specific object due to the use of a centralized data base and i.s now being developed further on the basis of using a unit system of software ~nd hardware of the SM EVM [International Small Computer] series. Bibliography BIBLIOGRAPHY 1. "Informatsionnyye sistemy obshchego naznacheniya" [General Purpose Information 5ystems], Moscow, Statistika, 1975. 2. Martin, J., "Organizatsiya baz dannykh v vychislitel'nykh sistemakh" [Organiza- tion o:: Data Bases in Computer SystemsJ, Moscow, Mir, 1978. 3. Vol'ter, B. V., V. K. Sheremet'yev, and A. D. Lesovik, "Standardi~stior, of Communications in the Software Structure of the ASU TP Monitoring System," PRIBORY I SISTEMY UPRAVLENIYA, No 1, 1975. ~ , 4. Sheremet'yev,~~�K. and H. Berger, "'Phe Structural Approach to Construction of " Software of Info_:nativ:~ Functions of the ASU of a Large-Capacity Installation," VCPROSY PROMYSHLENNOY KIBERNETIKI, No 48, 1976. COPYRIGHT: :zd~itel'stv~ "Mashinostroyeniye". "Pribory i sistemy upravleniya", 1981 6521 CS 0: 1863~155 32 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY i~IULTIPROCESSOR SYSTEMS STATE COMMISSION ACCEPTS INSTITUTE OF CYBERNETICS DESIGN b'OR MULTIPROCESSOR SUPERCOMPUTER Moscow SOTSIALISTICHESKAYA INDUSTRIYA in Russian 19 Feb 81 p 2 [Article by correspondent Zh. Tkachenko: "Computers and Systems"] [Text] A State Commission has accepted the design of a new, high-performance, multiprocessor computer. The computer is unlike any other in the world. T. Mar'yanovich, Party committee secretary at the Institute of Cybernetics of the Ukrainian Academy of Sciences, relates: "At the institute we have also created the latest automated systems for various levels." This concerns systems to control separate technological processes, enterprises, large territorial associations and sector complexes. During the past f ive years the introduction of developments made by this advanced group of workers has produced an economic effect of 195 million rubles. CSO: 1863/23-P 33 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR UFFICIAL USE ONLY UDC 518.74:007:57 CONSTRUCTION OF ALGORITHMIC LANGUAGE FAMILIES FOR PROGRAMMING AND DESIGNING MULTIPROCESSOR COMPUTER SYSTEMS Kiev KIBERNETIKA in Russian No 1, Jan-Feb 81 (signed to press 13 Feb 81) pp 1-7 [Article by Viktor rlikhaylovich Glushkov, academician and director of the UkSSR Academy of Sciences' Institute of Cybernetics, Kiev, Yuliya Vladimirovna Kapitonova, doctor of physical and mathematical science, laboratory director, Institute of Cybernetics of the UkSSR Academy of Sciences, Kiev, and Aleksandr Adol'fovich Letichevskiy, doctor of physical and mathematical science, section chief, Institute of Cybernetics of the UkSSR Academy of Sciences, Kiev. Received by the editors on 16 Nov 1980] [TextJ In articles [1] and [2], a method of formalized technical tasks for design of discrete systems was proposed. Use of this method for design of this particular class of systems requires development of mathematical models of this class of sys- tems, methods and facilities for solving the basic problems of design of analysis, synthesis and optimization, as well as language facilities for representation of the systems being designed at the various stages. A review of mathematical models and some methods for solving the basic problems of design of multiprocessor systems was made in article [3]. Here we shall consider the language aspects of design of multiprocessor system circuitry and software by the method of formalized technical tasks. The diversity of inethods and means used in the various stages of design leads to the need of organizing language facilities into a family of coordinated languages that have a varied functional purpose. Independent development of various types of language facilities (problem orienta- tion, functional and procedural languages, machine-independent and machine-oriented languages, etc.) prevails to a great extent in today's practice of programming. - Attempts to unify the diverse facilities and construct "large" languages such as PL/1, ALGOL-68, ADA and others, even though they achieve conveniences in some cases, have at the same time considerable shortcomings, not the least among which ar e the difficulties of efficient realization. Use of different algorithmic languages requires a coordination of them that would provide the capability of transfer from one language to another at different levels of realization, as well as the capability of using the programs written in the dif- ferent languages within one system. Coordination of languages of a family is pro- _ vided by the common character of the mathematical models behind their semantics, and the couunon requirements that express the relationship between the user, the - language and the system in which the language is realized and used. 34 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400024412-2 FOR OFFICiAi. USE ONI,Y This article is a generalization of the experience of the development and use of language facilities in the activities on computer software by the UkSSR Academy of Sciences' Institute of Cybernetics. It also contains the methodological principles . for developing algorithmic languages for new computer systems under development with the participation of the Institute of Cybernetics of the UkSSR Academy o� Sciences. Formation of the point of view presented in this article was influenced by the ideas of today's technology of programming clearly expressed in the "Works of the Working Conference of the IFIP [International Federation of Information Processing]" [4]. The PASCAL language was selected as the basis for traditional language constructions [5]. � Let us note that the idea of creating a family of coordinated languages for parallel programming is also discussed in article [13], although from somewhat different starting positions. This work may also be labeled a sufficiently complete review of today's facilities for parallel programming. Classification of Languages of a Family - Languages of a family are classified by several features. _ The first feature corresponds to program concept levels. The following four basic concept levels are distinguished: --problem; --functional; --sequential-algorithmic; and --parallel-algorithmic. The problem level is used to define the classes of objects ~zith which the program - operates and their properties and the formulation of the problem that the program solves. A problem program expresses a requirement of constructing objects meeting particular properties and in particular relations with source data. The functional level is used to define the functions computed by a program. At this level, recur- sive functional definitions are widely used. A functional program requires compu- tation of values of particular functions for particular values of arguments. The sequential-algorithmic concept level defines a program as a means of generating sequential processes of computations ~ith the information medium in which the program operates. The parallel-algorithmic level defines the aggregate of sequential programs or de- vices (modules) functioning synchronously or asynchronously, interacting among them- selves by exchange of data and control information (messages). - Facilities of different levels may be used in one and the same program. Therefore, the referral of a language to a particular level is not stringent and means that the facilities of a given level are most developed in the language in question and comprise its methodological basis. From the point of view of compatibility, it is convenient to consider any program as a program of the parallel-algorithmic level, isolating the other levels as special cases. A sequential program may be considered as a parallel program consisting of one module, and a functional or problem--as a . sequential program defining the single step process of computations (nondetermined, generally speaking, in the problem case). - 35 FnR OFFI('iAT. ?f,4F. (1Ni.Y APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R000400024412-2 - FOR OFF(C.I~~L USE ONI.Y The second feature for classification of language families involves the orientation to methods of realization. Let us distinguish the following basic methods of realizaCion: interpretation; translation; complexing of off- the-shelf modules; and hardware circuit implementation. Languages may be classified by orientation to problem classes. This classification may be rough (numerical programs, string processing, data base management, etc.) or more refined (problems ~f linear algebra, optimal control, translation, designing of specific classes of objects, etc.). Finally, an important feature of classification is the orientation to the user, for example: the applications programmer; the systems programmer; and the hardware developer. _ Data Types Scalar types are defined just as in the PASCAL language. They include logic (bit) values, number and symbol sets and finite sets of objects that are specif ied by a listing of the designations of the elements of these sets. Additional scalar types may be included in system programming languages to dPSignate parts of machine words (byte, halfword, etc.), addresses, names, etc. Structural types, in the final analysis, are made up of scalar and are divided into types of strings and f iles and the language and theoretical-set types of the compo- nent objects. Structural types are also divided into static and dynamic. All static type objects have identical dimensions and the value of each variab le of the static type may be stored in some previously assigned, connected region of storage. The dimensions of this region are defined at the time of the program call and may depend on the values of the actual parameters of this program. On the other hand, dynamic type obj ects have no fixed dimensions, and the memory for storage of values of a dynamic variable must be assigned in the process of program operation. In addition to these types, there are special structural types in a language for input- output operat~ons: input and output files. Strings are sequences of s}rmbols or bit strings with a fixed dimension. Static type strings have a f ixed length. The length of dynamic type strings may vary. Basic operations on static type strings allow only symbol-by-symbol processing, while complicated operations such as paCtern recognition and substitution may be made on dynamic strings. Arrays. The concept of an array as used here gener3lizes the tr3ditional concept of a rectangular array with integral subscripts. An array is a mapping of a set C, called the domain of arrangment, into the domain of data D of a particular type. _ The range of definition of an array A~C ~ D (the sign of inclusion indicates - that A is a partial mapping from C into D) is always assumed as finite. The array - type is defined by the D type of the values of its elements, the domain of arrange- ment of C and the set C~ ~ C, which limits the range of definition of this array. 36 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 . i FOR OFF[CI~L USE ONLY The domain of arrangement is usually a Cartesian product C= C1 X...X Cm scalar types or types of strings, and C~ = Ci~~ X...X Cmo~ is the Cartesian product of ty-pes contained in C1, C~. The number m is the dimension of the array. A conventional rectangular array is obtained if C1 = C~ = Z is a set of integers, and C~~~, C~~) are f init e intervals. The finite set CQ is neces- sarily indicated for static arrays. Fo r dynamic arrays, Cp = C. The dynamic array concept is usually applied f or representation of large volumes of data that may be arranged in external storage (on disks) . Access to array elements is determined, as usual, using variables with subscripts. However, in additic~n to operations on e lements of arrays, operations on arrays may be performed. Taken as the basis is tl~e set of operations defined in [6] and which includes elminations, overlay, shifts and element-by-element operations. - - Component objects are used to represen t heterogeneous data. The static component corresponds to the c4ncept of notation in the PASCAL language. Dynamic components are def ined in articles [7] and [8]. The concept of a dynamic component has a number of advani:ages ~ver the corresponding facilities in the ALGOL-68, PASCAL and PL/1 languages. First, it makes it possible to operate with components con- sidered as a single whole, and not only with elements of the components. Second, the concept of a component is independ ent of representation and may be considered an abstract type. Let us consider the definition of a component object. Let SZ be some set of elements called marks. (Arnost`)--a nonnegative integer--is specified for some marks. I~a.rks of (arnost') 0 are used to desion ate primary components and may, generally - s~eaking, include any scalar types of d ata. A component ob~ect is an oriented - multigraph with an isolated initial ve r tex, in the set of vertices of which Q has been assigned the function of marks cc, which maps the set Q into SZ. If the mark a(q) of the vertex q has a positive (arnost') m, then m arrows come out from this vertex and they are all numbered with d~.fferent natural numbers from 1 to m. If the (arnost' ) of the mark is not specif ied, then the number of arrows may be arbi- trary. All vertices of the component must be accessible from the initial. Also - discussed in [8] are (klubki) of component ob~ects. A set of components is called a(klubok) if together with each objec t it contains all its subobjects, and if the initial vertices of two components coin cide, then these componenf:s also coincide. ~iarks of components are consisered ope r ations for forming complex objects. The algebra of components, which is obtain e d with a consideration of these operations, _ forma the basis for defining computatio ns components. It is also used to define the external language represent a tion of component objects. Language ObjectG. =,:,r external representation of data, first of all component objects anc~ :;~rings, data language is u sed. It is not fixed and various programs may be P~~usted to various data languag es through the use of special facilities Fo:. describing the syntax. Objects tha t allow description in data language are called language objects. Strings and c omponent objects are used for their internal - representation. Elements of algoritiimic languages of a family may be i.mbedded in the data language. This makes it poss ible to use languages of a family as lang~-ages ror de~ioning algorithms. 37 FOR OFFICI.~L USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850ROOQ4QOQ2Q012-2 FOR OFFICIAL USE ONLY Theoretical set structures are used, first of all, in the initial stages of development of algorithms, in particular, at the problem and functional levels. They include sets, relations and functi~ns constructed with the operations per- mitted in the language. Set elements may be any data structures already defined in the language. Theoretical set types of data are considered abstract types of ~ a special form. In the final analysis, theoretical set objects are realized that can be used in algorithms may be indicated at the intermediate level of development. The general methodology for introduction and use of theoretical set data structures is described in [9]. Information iied~um and Structure of the Program It is assumed that languages of a family are realized in a system that uses a specific organization of orograms and data. A prototyne of this organization is the organization of the PROYeKT system [lOJ and its realization the PROYeKT-YeS which was i.mplemented in recent years on machines in the Ye5 [unified system] series. Each object that is input into the system, be it a ~rogram or data struc- - ture, is input into the information medium of the system and changes its status, ~ahich is defined by the aggregate of the objects put into the system earlier and the links between these objects and the adjustable mechanisms f or processing and using them. isan and sy~tem interact with high-level languages which include: directive languages; data languages; and ~ algorithmic languages. ~11 objects of the information medium are divided into information libraries. Each inf ormation library contains the programs and data used in solving some problem or class of problems. In particular, an information library may correspond to a pack- age of application programs that use a certain form of organization or a system of programs for some applied field. Each information library may be adjusted to its own languages (directive, data and program). In addition, an infonnation library is adj usted to the particular types of obj ects that caay be in it . An information library is divided into sections, each of which uses its own system _ for na~ing ob~ects and the link between them. A section of an inf orm2ticn library consists of modules which are divided into basic and subordinate. Basic module nanes are called global. These names may be used directly by all modules of the section. Each basic section may have modules subordinate to it. These modules are given local names which are collected into the local nomenclatures of the basic module. Tr~TO basic modules may use comnon local nomenclatures if this is indicated when they are put into the system. Local nomenclatures may be divided by types of named ob~ects. External representations of names in these no~nenclatures may coincide. They are distinguished in texts by the use of names of types. Values of local names are directly accessible from a basic module and ~.oaules subordinate to it. Linkage between the various information libraries and sections can organized by - the use of complex names (name of tt!e inforrsation library, name of the sc:.~ ~cn, name of the basic module, type and local name). In the process, it must be kepc: ~.~t mind that calls to other basic modules, sections or information libraries are more costly than calls within one basic module. In addition, such calls must ~eet cer- tain limitations imposed by the diversity of languages and nomenclatures used. ~Sodules are divided into program and data modules. The external representation of a program module consists of a title, descriptions and a statement. Indicated in - the title are the formal parameters used in calling a given module from others, and information on subor3ination. Some descriptions of a basic module are common to 38 FOR OF'FIC[AL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850ROOQ4QOQ2Q012-2 FOR OFFICIAL USE ONLl' all modules subordinate to it (exactly what kind is determined by the programming languages uged). Problem Level Descriptions at the problem level have the nature of mathematical def initions. Defined are the necessary types of data, among which the basic role is played by abstract and theoretical set types, objects, variables, functions and subroutines. Descriptions at this level may be incomplete; for a function, for example, perhaps only its designation, type of arguments and type of value are indicated. To con- - struct objects and formulate conditions, theoretical set constructions, quantifiers, - recursive definitions and other facilities of the language of mathematics are used extensively. As the basis for formalization of these facilities, the facilities of = the language of the theory of a system for processing mathematical texts may be used [11] and [12]. A reasonably restricted language of the pro~rlem level may be realized directly through translation, interpre~ation or of programs from off-the-shelf modules. The capabilities of parallel data processing in a _ multiprocessor system are derived in the process directly from the definiCion of the conditions which the obj ects being designed must meet. In particular, parallel _ computations permit theoretical set operations, checking of conditions that contain quantifiers, etc. Richer languages are used to formulate the initial technical tasks which in turn are used as data for design programs. These programs make it possiUle to effect suc- cessive ref inement of the algorithm being designed through construction of inter- mediate realizations of types of data, statements, functions and subroutines, to perform optimizing transformations, and check the dynamic and static properties of the program being designed. Functional Level _ The basic facilities of this level are recursive functional definitions which may be represented by systems of equations of the form: fi(xl, Xn) = Fi~fl~ fm, xl, Xn)~ i= 1, m, where fl, f~ are unknown (determinable) functions, and Fi are expressions constructed from sycr.bols of the unknawn functions, arguments xl, xn and constants. The smallest fixed point is considered the solution. To calculate t~e value of f i(al, an), each equation is developed into an infinite tree and analysis of computations occurs si.multaneously along all branches with the specified set of values of the arguments. If the width of the tree is sufficiently large, ~ then parallel computations may be organized. Especially important is the case when the unknown functions fl, fm are arrays, and the sets of values of the argu- ments (xl, xn) traverse some subset E[ C of ~he domain of arrangement of C. For example, let there be a need t~ compute the value of the arrays fl, f~ at all points of the domain E. The possibility of parallel computations is now deter- mined by the width of the front of the wave F.k ~ E, which consists of those sets (xl, xn) ~ E, for which the value of all functions fl, fm in the k step can be simultaneously computed. If the front of the wave is sufficiently large, then parallel computations can again be organ3zed. As applied to the arrays, the system of functional equations may be brought to the structural form: 39 FOR i: FFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY fi = Fi (fl, fm)~ i= 1, m, where Fi are expressions in the algebra of the arrays [6]. T~he functional level also includes facilities for transfo.rnation of component objects and strings using systems of relations. Called a relation is the equality of the foYm F(xl, xn) = G(xl, xn), where F and G are expressions in the algebra of the component objects, and xl, xn are parameters. This�relati~n is said to apply to the component object S, if there exist obj ects al, an such that S= F(al, an), and the object G(al, a~) is called ir. the process the result of the application of this relation to S. The equality of the components in the process may indicate either their isomorphism or equivalence. In the ge~reral case, the relation may,be supplied with the condition P(xl, xn), which the sought objects al, an must meet. If it is necessary to apply a system of re- ' lations to all subobjects of some component, then parallelism may be used botYi to check applicability and to apply the relations simulataneously to several subob- - jects. The technique of applying relations was developed in the ANALITIK language (14] and the programming languages of the PROYeKT system [10]. Problems of parallel realization of relations were studied in article [15]. The facilities of the functional level may be used in combination with the facili- - ties of both the problem and the sequential-algorithmic levels. In the latter case, ad3itional possibilities of parallel processing emerge through use of the overlap technique [3] for the sequentially executed parallel statements. ~ Sequential- and Parallel-Algorithmic Level The basic facilities of the sequential-algorithmic level do not differ fundamentally from the standard facilities of traditional programming languages, except that use of certain types of parallel statements (parallel or group assignsnent, for example) is permitted. Let us examine the parallel-algorithmic facilities. A mathematical model of a parallel program is a multilevel network of algorithmic modules with a variable structure [3]. The concept of an algorirhmic module is derived through combination of two concepts: of a program that describes the functioning of a module, and of a - component [komponenta]--of a location where the given program is executed. In the same component, different programs may be executed at different times. Descriptions of the components and the programs executed in them are placed among the descrip- tions of the parallel program which itself represents control of this network and is called a control program. Programs situated within the control are called sub- ordinate with respect to their own control program. Since multilevel control is permitted, subordinate programs may be controlling with respect to programs at a Iower level. A program is combined with a component by a control program by using the initializa- � tion statement VYPOLNIT' PROGRAMMU A(xl, x~) V KOMPONENTE K.[execute program A(x~, in component K]. If component K is free at the given time, then - 40 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFI~'IAI. USE ONLY the program A is put into this component (or initialized in it) with the actual parameters x1, m. If another pragram is being executed in K, then the state- ment A(xl, xm) goes to the queue and will be executed in K after completion of the program already in it and all the programs that are already in the queue. If only one program is always executed in some component, then the description of this component and the program may be combined in one description. Such a compo- nent begins operating automatically simultaneously with the call of the control program. Interaction of Components. To describe interaction of components in the process of their parallel operation, two basic mechanisms are used: direct exchange of infor- mation and us e of common memory. Direct exchange of information occurs with the use of static and dynamic links between the components.~ Static li.nks are described just as in the ALGORITM language [10]--through indication of identification between input and output vari~b les 1 (channels of the components). Descrj.ptions of these variables are included in the descriptions of the components. An input (output) variable of a component must be ~ assigned tihe,:,s~ec~.fic type of data that is transmitted through the corresponding channel. Iri`~'addition, these variables are assigned specific properties that describe the data transmission method (control variables, informational, with buf- fering, queues, etc.). Descriptions of input and output variables, j ust as the other descriptions placed in the description of the component, may be used in the programs executed in the given component. To facilitate automatic inclus ion of these descriptions in a program, components having an identical description, are assigned a type. This type may be described separately, and its name may be used to simplify descriptions of components and corresponding programs. If the type of program does not correspond to the type of component in which it must be executed (for example, the program has its own input and output variables), then the corres- pondence between them is established by including additional information in the initialization statement. Dynamic links are not explicitly described, but are formed during execution of ex- change statements. Two types of exchange are differentiated: transmission of messages and exchange of data. Messages are transmitted using the statement: PEREDAT' SOOBSHCHENIYE X PROGRAMMEE A(K) [send message X to program A(K)], where - A is the program name and K is the component name. During execution of this state- ment, a dynamic link is established between the given component and component K, and message X is sent through the corresponding channel and written in the message buf- fer for component K. Access to messages received for the given program A is possi- ble only during operation of this program and is realized by using the statement: ~ OBRABOTAT' SOOBSHCHENIYE [process message]. ~ ' The structure of inessages which components exchange betw~er~ themselves is described in the control program and is common to all components and programs subord inate to it both directly and at lower levels of control. Data is exchanged by using two statements: PEREDAT' DANNYYE X PROGRANIME A(K) [send data X to program A(K)] and PRINYAT' DANNYYE Y IZ PROGRAMMY B(M) [receive data Y from program B(M)]. These two statements correspond to each other. Estab- ' lishment of a link and transmission of data X from component K t~ component M wi1l . 41 FnR nF~rrr,?~ rrc~ nu~ v APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY occur at the moment when simultaneously, program B executes the corresponding data transmission statement in component M and program A executes the receive statement in component K. Data X is written in area Y. Before this moment, execution of each of these statements is in the wait state. To reduce the waiting period, the apparatus of inessage transmission with interrupts may be used. The corresponding statement has the form: PEREDAT' SOOBSHCHENIYE X PROGRAA'Il~4E A(K) S PRERYVANIYEM ~ [send message X to program A(K) with interrupt]. Execution of it halts the opera- tion of component K and initiates execution of the interrupt program, which in this case must be described in program A. If the gap between execution of the exchange statements is expected ~o be long, then the components may be synchronized between themselves, exchanging messages on readiness to send and receive data. - - Facilities for interaction of parallel processes are easily expressed in terms of the dynamic interaction of components in such models as, for example, Hansen's model [16], which we believe c~osest to o,ur considerations. Let us note that interaction with the use of dynamic links corresponds more to program realization, while static links are used more extensively in designing hardware. The combined use of both facilities is the basis for joint design of computer circuitry and software. Common memory for components of one level is def ined by descriptions of those - variables that have the special OBSHCHAYA [common] feature. In addition, data from an information library is accessible to a11 programs. A call to common memory occurs during execution of a statement of assignment, checking conditions in condi- ~ional statements, etc. Conflict calls are queued. In addition, the OTKRYT' [open] and ZAKRYVAT' [close] statements make it possible for some program to seize certain sections of common memory for a time. Partial se.zure is also possible, for exam- ple, ZAKRYT' DLYA ZAPISI (CHTENIYA) [close for writing (reading)]. Control. Description of a large number of homogeneous components may be facilita- _ ted by use of arrays of components, the dimensions of which may depend on the _ actual param~ters of the control program. Fully decentralized~ functioning of com- ponents corresponds to the dummy statement part of the control program. In this - case, each component must be described along with its program and all components begin operating simultaneously at the time of the call of the control program. A program ends its operation at the time when all components end operation. In the case of the nontrivial statement part, a control program may execute initi- alizaCion of operation of components, interrupt their operation, track the end of ~ operation of components, check cunditions of their operation, etc. In addition, a control program may change the structure of links between components, and eliminate and introduce new components from a previously described reserve. St:.].1. another facility for dynamic change of structure involves the dynamic call of programs. Any program maq call any other program described at the same or higher level of the hierarchy. In some cases, a call is permitted with lower levels too. Dynamic calls may occur recursively. With a call of a parallel program from con- tro1, a change in the structure of the components occurs through addition of new components which are eliminated when the called program ends operation. Use of the apparatus of checkpoints makes it possible to organize interaction of coprograms. 42 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFI~.IAL USE ON1.Y Conclusion This article presents approaches to constructing a family of coordinated algorith- miC languages for design of multiprocessor systema circuitry and software, which are - under development at the Institute of Cybernetics of the UkSSR Academy of Sciences. Several representatives of this family have now been developed. They include: the ALGORITM-80 language oriented to joint design of hardware and software, the MMrP multimodular programming language that belongs to the parallel-algorithmic level, and the AMK arithmetic macroconveyor language pertaining to the problem-functional level and oriented to the problems of computer mathematics. The latter two languages are intended for programming recursive computers [17]. References 1. Glushkov, V. M.; Kapitonova, Yu. V.; and Letichevskiy, A. A., "Theoretical Principles for Design of Discrete Systems," KIBERNETIKA, No 6, 1977, pp 5-20. 2. Glushkov, V. M.; Kapitonova, Yu. V.; and Letichevskiy, A. A., "Use of Methods of Formalized Technical Tasks for Design of Programs for Processing Data Structures," PROGR~fIROVANIYE, No 6, 197$, pp 31-43. 3. Glush~o~;"'V. M.; Kapitonova, Yu. V.; and Letichevskiy, A. A., "Theory of Design of Circuitry and Software for Multiprocessor Computers," KIBEitNETIKA, No 6, 1978, pp 1-15. - 4. "Design of Reliable Software," in "Trudy rabochey konferentsii IFIP" [Works of the Working Conference of the International Federation of Information Process- ing], Novosibirsk, VTs SO AN SSSR [Computing Center of the Siberian Branch of the USSR Academy of SciencesJ, 1977. 5. Wir*h, N. "Sistematicheskoye programmirovaniye. Vvedeniye" [Systematic Program- mii.ig: An Introduction], Moscow, Mir, 1976, 183 pages. 6. Glushkov, V. M.; Kapitonova, Yu. V.; and Letichevskiy, A. A., "Data Structure Theory and Synchronous Parallel Computations," KIBERNETIKA, No 6, 1976, pp 2-15. 7. Gorokhovskiy, S. S.; Kapitonova, Yu. V.; and Letichevskiy, A. A., "Programming Facilities and Solving Logic Problems in Software Systems, Basic Concepts of the [word illegible] Language," KIBERNETIKA, No 3, 1974, pp 27-47. 8. Glushkov, V. M. and others, "Basic Instrumental Language of Programming," (Preprint of Institute of Cybernetics of the UkSSR Academy of Sciences, No 79-22), Kiev, 1979. "Programming in the L2B Language," (Preprint of the Institute of Cybernetics of the UkSSR Academy of Sciences, No 79-23), Kiev, 1979. 9. Glushkov, V. M.; Kapitonova, Yu. V.; and Letichevskiy, A. A., "Instrumental Facilities for Design of Programs for Processing of Mathematical Texts," KIBERNETIKA, No 2, 1979, pp 37-42. 10. Glushkov, V. M.; Kapitonova, Yu. V.; and Letichevskiy, A. A., "Avtomatizatsiya proyektirovaniya vychislitel'nykh mashin" [Automation of Design of Computers], Kiev, Naukova dumka, 1975, 232 pages. 11. Glushkov, V. M.; Kapitonova, Yu. V.; Letichevskiy, A. A.; Vershinin, K. P.; and Malevanyy, N. L., "Constructing a Practical Formal Language for Writing Mathematical Theories," KIBERNETIKA, No 5, 1972, pp 19-28. 43 FnR (1FFTrt e r r rcF nNt v APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR ()FFICIAL USE ONLY 12. Glushkov, V. M. "System for Automation of Proofs (SAD)," in the book "Avtomatizatsiya obrabotki matematicheskikh tekstov" [Automation of Processing of Mathematical Texts], Kiev, Institute of Cybernetics of the UkSSR Academy of Sciences, 1980, pp 3-30. 13. Kotov, V. "parallel Programming Laaguages," Part 1, KIBERNETIKA, No 3, pp 1-12; Part 2, KIBERNETIKA, No 4, 1980, pp 1-10. 14. Glushkov, p, M.; Bodnarchuk, V. G.; Grinchenko, T. A. and others, "ANALITIK (Algorithmic Language for Description of Computer Processes with the Use of Analytic Transformations)," KIBERNETIKA, No 3, 1971, pp 102-134. 15. Shevchenko, V, p, "Parallel Computations in the Algebra of Component Objects," "Automated Extracts of Dissertations of Candidates of Physicomathematical Science," Kiev, KGU [Kiev State Universi~y imeni T. G. Shevchenko, Order of Lenin], 198Q, 20 pages. 16. Hansen, p, g.~ ~~Distributed Processes: A Concurrent Programming Concept," CACM [Communications of the Association for Computing Machinery], Vol 21, No 11, 1978~ pp 934-941. 17. Glushkov, V. M. and others, "Recursive Machines and Computing Technology," IF~P Congress-74," Stockholm, 1974, pp 65-70. CQPYRIGHT: IZDATEL'ST~JO "NAUKOVA DUMKA", "KIBERNETIKA", 1981 8545 CSO: 1863/149 - 44 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R000400024412-2 FOR OFFICIAL USE ONLY VOICE RECOGNITION/SYNTHESIS MIr1I-COMPUTER VOI CE COMMUNI CATION SYST~I Moscow PRIBORY I SISTLMY UPRAVLIIVIYA in Russian No 1~ Jan 81 pp 13-11~ [Article by engineers V.K. Kucherenko and I.1t. Kucherenko: ~~Mini-computer Speech Communication System"] " [Text] This article examines an experimental system of voice communication ~rith mini-computer~ of aubaystems for segmented recognition of speech forms and the segmented synthesis of speech transmisaions. The speech communi.cation system is SM-1 mini-computer based and operates real time. A feature of the system described is the uae of interval durations between nulls in the speech signa].~ and also between adjacent points of local extremes of the speech signal. _ l~le ~11 review several general questions arising in the development of the computer based speech communication systems~ specifically a number of limitations which are being de-bugged for subsystems(routines) involving apeech form recognition and also synthesis of speech communications~ ereated for pratical application. The most substantial limitation is associated with the recognition time for speech form and synthesis time for speech communications~ both of which must be rather low i.n order to realize one of the basic advantages of speech communication: convenience and speed of data transmission from the user to the computer and conversely. Another limitation for t,he speech form recognition routine results from require- ments imposed on pror.unciation conditions for voice communications and speakers [1]. Necessary requirements for the practical appZication of speech communication with computers in several small systems include also low cost of the speech communication _ system and small size. Idaturally~ a sufficiently flexi.ble and powerful computer system with the use of speech communications must permit simultaneous interface with several independent users working at their own terminal s. One of the ways to effectively solve this problem linked with the continual drop in the cost o~ mini-computers i~ to supply each terminal with a micro-computer~ whose functions include analysis and synthesis of the voice signal. 45 ~ FOR OFI~~iCIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY t4oreover, in small local systems not linked to powerf`tz7. computers~ the use of speech communications systems is possible with micro or mini computers. However~ _ use of the latter in developing efficient speech communications aystems with computers is difficult~ a situation explained by the relatively low rate of operations in the inexpensive micro oomputers and their low-volume current memories~ which result in a cor.siderable limitation of the reco~nition algorithm~s computing efficiency and of the entire set of speech communications programs. Thus, posing the task of developing speech communicai:ions systems computers for practical application leads specifically to the problem of bui.lding these micro and mini computer based systems. One of the important features of the speech communication system also is the capability for the current variation and expansion of the differentiated and synthesized speech communications compositions. - The article further examines the e~erimental speech communications systems with the mini computer developed by the authors~ easily incorporating various types of mini and micro computers and operating real time with a high speed capacity of 150-200~000 operations per second for the processor, The system described utilizes a SM-1 mini computer and an "Yelektroni_ka-60~~ micro computer. In utlizing the miero computer~ all archive a.~d program sets forming the mininal configuration speech communication syatem are situated in the micro computer's current storage. Minimum memory capacity necessary for speech form recognition routines is 4K 16-bit words~ which permite the recognition of several dozen commands. The minimum memory volume required for the apeech communications synthesis routine is dependent upon the vocabulary and prnnunciation quality of ~ speech and amounts to ~---21.K. In developing the micro ~?Yelektronika-60~~ computer based communication system~ a series of problems had to be resolved which were associated ~th regeneration of the c~ymamic semi-eonductor memory~ which influences the quality of synthesized speech sounds. Common to the speech communication recognition and synthesia routines axe the nardware options developed by the authors for the input of the speech signal to the computer and output from the computer. After input to the computer memory, the speech signal undergoes a preliminary processing stage~ which includes a logical averaging and a hysterisis filter:ing. [2]. Primary speech signal parameters in recognition of speech forms are form characteristics of histograms for i.nterval duration distribution between null intersections and adj acent exteme periods o� the speech signal i.n the analysis segment. To reduce the dimensions of the span for information c,haracters and further segmentation, a claster analysis is employed [31 and the evaluationa obtained by t~e authors relating to exactness of approximations of apeech signal structure frequency. Tile speech form recognition routine operates in .dialogue mode in real time during instruction as well as in recognition and prov~.des for verification of speakers and to differentiate with a high degree of reliability such similar sounding words as "sinus",,"minus"~ "Minkus"~ etc.~ where instruction in majority of instances is satisfied by a single realization of the speech form. iti'e will the speech communication synthesis, which provi.des for realizing any phonetic aystem in the Russian language. The speech communications aegmented synthesis routine includes~ in addition hardware and software for input-output of 46 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-00850R000400024412-2 FOR OFFIC[AL USE ONLY of the speech signal,a selection of programa for automated segmentation of the voice signal With the capability for subseqent control oi' segmentation results and a set of programs providing for the creation and editing of the segment dictionary~ and also aearch~ selection~ concatenation~ and the raproduction of dictionaxy elements in the required sequence. Ir. formulating the elements of the segment dictionary9 the voice material used to obtai.n the segment dictionaxy is subjected to automated segmentation with the claster analysis employed. The limita of the segments obtained during automatic segmentation are verified, if necessary~ based upon graphic and tabular data on spectral and time characteristics of the speech signal. The control of the _ natural aural quality of the isolated segments and their intelligibility is achieved by monitoring the segments via a loudspeaker. The voice segments thus obtained axe then used in constituting the segment dictionary. Formation of the segment di ctionaxy and synthesis of voice communications is effected the use of control programs~ the flow-chart for which is depicted in illustration 1. The control program organizes the dialogue between the user and the voice communica~,;ion synthesis During the dialogue process~ the control program provides t.he user with the capability to create or edit the the segment dictionary, and to organize the input of written communication, its automatic analysis and tran slation into the aural communication. The control program realizes its functions through the "Synthesis" routine and the "Archive~~ the flow-chart for which are depicted for one version of realizing the speech synthesis routines in illustrations 2 and 3 respectively. Navono ~1~ . enc ~2 Havano ' aync (3 rrccr 3~ en,~yx~e" (5) . /+em ~5) nct Aa ~PC� ' em 1 � ecc,r(6) ~ ~ g ~ A9!IC C6C (C~ ~ CPC? Qo B~�CuNme~~ . I lo IIFC , IeoHeq Hem (9, ~ 11'~ a~_, ~ Qa H 3 a~ (1 5) - Hpm~~3~ 6~ er~6 KoNeu (14) ~ ~ ' Illustration 1. I'low-chart for speech communications synthesis: 1. Begin 2. New segment dictionaxy: dump of Main routine ~'Archive~~ axid "Synthesis" respectively 3. Yes ;~tain routine ~'~A.r~}~iue" ::5. no 6. .input of the segment clictionary archive and its catalogue; 7. Synthesis of speech communication; 3. yes ~9. no �10.. ~Main routine ~~Synthesis~' 11. diction~uy editing 12. yes 13. no 14. end 47 ~ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL L1SE ONLY Illustra~tion ti: "Synthesis" Routine Flow-chaxt l. Begin 2. Input of written communication 3. isolation of the sequential segment of t~ie written communication 4. end of written commuriication~ 5. yea 6. reproduction of the synthesized speech communication 7. no 8. analysis of the current segment of the written communication and determination of the beginning of the next 9. collection of statistics on the segment dictionary 10. search for the required voice segment in the segment catalogue 11. end 12. segment located 13. no 14. diagnostic communication 15. yes 16. aegment selection and output btiffer loading Havano /1 ~ ~ ~ 21 HCf? em ,qo6C ~ em ftm ~ P m o ~ Q� C, Q n a em HR BNC DCN ~HC l)K? a RCR m _ �H? EN? �N? em CIf7 ~ EC K ' 1' em Qa ,qa m ~ BC CHN CC R Neu ' OX ON OK QC ~ Illus~tration 3: "Archive'~ Routine ~'~'low-chart l. Begin 2. new segment dictionary 3. no 4. yes 5. addition of segment o. segment re-designation 7. segment removal 8. segment re-distribution in the archive 9. segment re-distribution based upon statietics on the segment dictionary 10. intiali:ze segment dictionary catalogue 11. input of segment name 1.2. input of old segment name 13. input of segment name 1/~. catalogue print l~. print of catalogue contents 16. is there such a name in the catalogue? 17. dictionary and catalogue output lg. input of aegment dictionary archive and its cata.logue 19. segment input 20. input of new segment name 21. deletion of segment and compression of segment archive(file) 22. end 23. catalogue update 24. diagnostic communication Ttie "Synthesis~~ routine is caxried out by the input of the written communication~ its analysis~ search, selection of the appropriate speech segments~ their con- catenation, and also the loading of the output buffer with subsequent reproduction of its content vla the speaker. If the r.equired speech segment for the routine~ is not located, the diagnostic ~ommunication is output with th.e ~,i,:~sing segment indicated. In this case~ the user may add the necessary speech segment to the ~e~ment dictionary. At the end of its o~eration~ the ~~Synthesis~' routine assembles statistics of the freqency of the segment dictionary u sage frequency. The utility of such information lies in that the segment di ctionaxy is organized as a linear listing i, c~jmamic distribution in memory~ and there- fore may shorten the time for selecting elements of the dictionary, optimizing trieir relative locations. 48 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY 2'he "Archive" routine provides for the creation of a segment dictionary~ nnd also edits it through i.nput~ redesignation~ or elemination of segment dictionary elements. The capability for the output of current information on the segment dictionary catalogue~ and also of diagnostic communicationa in the event of error location prov.i.dea certain conveniencea in operations with the speech synthesis routine. _ The modular principle of construction for softwaxe on the routines for speech form recognition and speech communication synthesis permits creation of an easily modified and expanded aystem equipped ~rith mini or mi.cro computers for speech communication synthesis~ upon the needs of ;.he specific application. 41e urlll cover certain possible areas of use far the speech communications systems described above. Primasily~ this automated data-S.nformation system and control s.ystems of varying types are clistingui.shed by both the number of recognized and synthe~sized~ speech comr~unications and by demanda or reqtti.rements for computer equipped speech communication systemse The employmerit of speech communications makes possible the communication or interface the ASU or the data-information aystem and in~ormation to a broad conti.ngent of individuala unfamiliar with computers. Here the speech communication may also serve to verify authorized access to information. 'Phe sinplest in terms of speech communication with computers are the systems in which the voice material consists of a limited number of discrete communications. Speech communication with computers ma,y be employed also in ASU TP~ for AI~M servicing~ etc. Undoubtedly~ speech communication with computer is an effective approach in the construction of eronomically optimum man-machine systems. Literature l. Martin~ T. Practical Application of Speech Input To Computers. TIIER~ April, 1976, v. 64, rro 4. 2. Tu. Dzh., Gonsales R. Principles in Form Recognition~ M.: Mir~ 1978. - 3. Kucherenko, V.K.~ Kucherenko I.K. Acoustical Isolation of Homogeneous Segments in Speech Signals.--From the book: Problema in S~eech Comprehension Systems. M.: 1980. COPYRI.GHT: Izdatel~ stvo ~~Mashi.nostroyeniye". "Pribory i sistemy _ upravl eniya" ~ 1981 g~51 cso: 1s63/14z 49 - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-00850R040440020012-2 rux vrri~iAL U~~ UNLY NETWORKS UDC 681.324 - CONCEPT OF A MODERN COMPUTING NETWO}ZK Riga AVTOMATIKA I VYCHISLITEL'NAXA TEHI~NIKA in Russian No 2, Mar-Apr 3I (manuscript r.eceived S Dec 80) pp 3-14 [Article by ~.A. Yakubaytis] [Text] Computing networks are the highly efficient base of the moderu data pro- cessing industry. The first stage of their development was the creation of centralized tree networks in which n user stations were connected to a single computer via communications channels. At the second stage the ability to join a - small number of trees together was evidenced. And, finally, at the third, modern, stage distributed computing networks are baing created whose topology is determined by the efficiency of the performance of all data processing processes. ~ The first large computing network in which the concept of distributed computing and data processing resources was well develo~?ed was the ARPA network [1J. A further d evelopment of this concept was reflecte~? in the CYCLADES and CNET networks [1]. And, finally, appeared the DATAPAC, TELENET and TRANSPAC networks [1] in which the parameters and procedures for the excha.nge of information between computing com- plexes were standardized for the first time on the international level. In the - Soviet Union distributed networks werF. created in Moscow [2J, Novosibirsk [3], Riga [4] and a number of other scier_ce centers. An analysis of these generally different developments has demonstrated that they have a common basis which defines the concept ~f a modern distributed computing network. This article is devoted - to a discussion of this concept. l. Model ~ J The fundamental logical concept of a model of a computing network is the system-- a hierarchical group of functions im~;lemented in one or more computers, designed for solving specif ic network problems. Systes~s are linked to one another by means of conditiona,l lines, called physical connections, forming the logical structure of the computing network. Eack? system is open~ended, i.e., satisfies the general requirements [5) of the interaction architecture o~ open-ended systems of the International Organization for Standardization (MOS [TSO]). All systems are divided (cf. fig 1) into a number of layers, cal.led levels. Each - level performs in the computing networ.k a specific functional task and provides service for the 1eve1 above it. The digest of rules for the interaction of 50 FOR OFFICIA~. USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY entities o� like (identical) levels of various syste~s is called the protocol. _ The rules for the~ interaction o~ en~ities of ad,jacent leye~.s o~ on~ and the same system are cal7,ed the interlevel inter~ace. fucmer.+a A Cucma7+a B Cucmrna W ooou 1~ o000 00 _ ..i_.1_. .~.t. t. 7 - +-~t~ ~r~T~ ._._2)._._. ~ ' ~ _ I ~ Nrl ypoBeHe ~T~ ~ t. . .r.~... N . .Benr, 'T'~. .1..~ ._.yv~ .L. . ~ ~ ~ N-f oBeNb ~ T ..r.+. . .L..~. ._yP . L.a. . _ ..~.T. , I L--- L-----------~ J ~ 3?----------------- ~~u+,~cKUe c{~arna?a coeJeNei:uo ~r�~m~x cucm~+ _ Figure 1. Interrelationship of Applied Programs in a Computing Network Conventional symbols: _ 0 Applied program Applied program line of interaction Key: - 1. System A 3. Physical facilities �or connecting 2. Level open-ended systems ~ In the model ~f a computing network discussed by ISO (cf. fig 2) there are seven functional (logical) levels. The tasks per~ormed by these levels are described in table 1. The top two levels of the system define the processes (the processes of representing data and of controlling applied processes and admintstrative control). The bottom ~ive levels form the network method of access to these processes. The linking points between the network method of access and processes we will call ports. Table 1. Functions Pezkoxmed by System I,eve7.s No of Name o~ level k'unctions implez~ented by leVel level 7 Applied ~xecution of appl,ied programs o.f users, control of terminals and operator stations, adm:Cnistrative con- trol o~ network. [Continued on ~ollowing page] 51 FOR OFFICIA~, USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 ~ rux urt+~t,~A.L U5~ ONLY 6 Representation Representation (tnxezpretaxi.on) o~ ~he p~eaning . (vaJ,ue) .o~ informatfqn txans~ti~ted be.tweer~ a~plied programs, including eonveraion o~ ~n ..,uc~ions and data. 5 Contact Organization and pex~oxmance o~ co~uunications contact between applied programs. 4 Transport Trans~er between processes of data ~i1es coded in any manner. 3 Network Routing and addressing ox in~ormation; contzol of data array flows. . 2 Channel Establishing, maintaining and breaking a connection. 1 Physical Physical, mechanical and functional characteristics of connection. 'r'y//.1'((UONQAyI/~e ypoBNU: 7 /Ipw,bdiiae~ ~ - , 710) G /~xdcmQBumene++eiti Q` ~ _ _ - /In~m~ S CeoNta9u~i ~ 11) 4 Tponrnoo~,,,,xi F~i 3 3 Ceme6wi � ~o ~ ~ ~ ? KnMZneNaiu ~ 12 ~ ` ~ _ " u 1 4~u)uvPCK~[~ a 9 ) ~ ~ < 9~u~uveeKOC 1 3 ~ ~ I coeduNetive _ Figure 2. Hierarchy o~ System I,evels Key: 1. Physical 8. ~'unctional levels 2. Channel 9. Logic channel,s 3. Network 10. ~roceases - 4. Transport 11. 7?oxts 5. Contact 12. Network method of access 6. Representation 13. Physical connection 7. Applied 52 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OF'FIC[AL USE ONLY The solution of diverse netwoxk probJ.ems is made possible by rqean~ o~ the special- ization o~ systems, i1,].ustrated in table 2. Working, tex~ina~, and administrative " systems are ca~led user systems. A11 systems have a 7-1~eve1 stxucture. However, in administrative and communication systems ievele 4 to 7 axe uaed on7,y ~or ad- ministrative contxol. zn the remaining systema these leve~,s both f or basic (involving the interaction a~ users' pro~rams) and administratiye contxol. Table 2. Computing Network Systems No Name o� system Tasks performed by system ~ 1 Working Assignment of network resources: storage o~ data arrays, data search, execution o.f computing o~exations, simula- tion o~ processes, phenomena and entities, and develop- ment o~ so�tware. 2 Terminal Use of network resources: contrql o,f the operation of terminals, preparation o~ assignments, and interfacing with technological processes ~or the purpose of ineasuring and controlling them. 3 Con~munication Routing of flows of arrays of information transmitted between working or terminal systems. 4 Administrative Administrative control of computing network (gathering of statistics, operation records, issuing reports, diag- nosis o~ failures, information regarding operation of network, etc.). With the introduction of the specialization of systems it is not di~ficult to produce the logical structure shown in fig 3. Its nucleus is the logical szructure - of the data transmission network, consisting of communication systems l~.nked with one another by means of physical connections. After fulfillment of the require- ments for a standard interface, user (working, terminal and administrative) systems are connected to the data transmission network, as the result of which (cf. fig 3) the logical structure of the computing network is formed. User systems satiskying the standards for intex~acing with a data tza,nstAission network are connected directly to co~nunication systetns. .z~ a usex system does not conform to these standards, then it is connected to the couupunication system _ via a logic element o~ the neCwork called an inter,face modu~,e. When communications contact is perf oxmed, the ports o~ processes located in various systems are con- nected by means o~ conditional lines ca~led (c~. ~ig 2) logic channel.s. The trans- ~er o~ data from one process to another takes place through them, Data input/ou~put (W) logic modules l,inked with individua7, tezminal systems com- plete (c~. ~ig 3) the synthesis o~ the l.ogica~. structure o~ a compuzing network. - Computing networks are not always, as in fig 3, single-level. ~'hey can ~orm (c~, f ig 4) hierarchies in which any o~ user systems interact regaxdless of on which level og the hierarchy they are. ~ 53 FOR OFFICIA~. USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 ruK ua~r�~ac:lAL USE ONLY 2 ) Qhrn~r~um;arr~ cemb ~-i~ ~ ~ ~ ~ - ~ ~ Ccm~, 3BM / e ~~~3T~~ . / / \ \ ~ ~ ~ ~ ~'!/716 L~QN!! [~QHHb/Y ~ ` . ~ ~ \\1 4" i - I 7 ~ I I / ~ ~ ~ ~ ~ ~ ~ 1 \ ~ ~ ~ BA / 1 \ 1 ~ / ~ \ \ ~ _ \ \ l-JJ ~ ~ / \ \ ~ ~ ~ - i \ ~ i \ ` ?\B ~i \ ~,g ~ / Figure 3. Logical Structure of Computing Network Conventional symbols: User system ~ Communication system -Q?- Interface module Input/output module ~ Physical connection and point of standard interface Key: l. I/0 unit 3. Computer network 2. Computing network 4. Data transmission network Inter.face points at which communication systems are connected with user systems or interf~ce modules must satis~y an international standard called Recomttlendation X.25. It was introduced [6) by the Consultative Committee on ~ntexnational Tele- graphy and Telephony (CCITT) and has become widespread in many countries of the world. For example, in table 3 are given [7J some data on Stafie data transmission networks created or to be created in Europe. Recommendation X.25 ca11s for the collective use o~ data transmission channels by the creation o~ vixtual (logic) channels and by the transmission o~ standard ~ackets thxough the~. 54 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY 1) !l~n,~~ ucpnpxu~ 66N/uCAwnP~ibh6/X 3 ceme~ ~ ~ ;~a; - - _ . 1 1 , - - p ~ . ~ . - ~ _ ~ Figure 4. Hierarchy of Computing Networks Key: ~ - l. Levels of hierarchy of computing networks Table 3. European Networks for Transmission of X.25 Packets No Country Year of Number Nutabez of terminal Numbex o~ chaxacter-user entry o~ net- systetns in 7.980 stations in 1.980 , in to wo rk service centers ~ 1 England J.979 3 1.00 1.057 2 Belgium ~ 1980 3 250 7.50 3 Holland 1979 3 1400 0 4 Denmark 1981 1 , - [Continued on following pageJ 55 _ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 ruK urr~c:~~L USE ONLY 5 Spain 1979 S 250 90Q0 . 6 Italy 1981 3 - ~ 7 Norway 1981 1 - ~ - 8 Finland 1982 1 - ~ 9 France 1.978 25 3000 3000 _ 10 FRG 1979 7 >500 >1500 11 Sweden 1981 1 - A 7-level hierarchy of computing network protocols has been introduced in con~ormity witti system levels (c�. :~ig 2). These pzotocols (c,f. fig 5) haYe the same names as the levels do. Generally several different protocols can be at the same level. ~ The protocols oE di.fferent levels are quasi-independent of one another. Therefore, the replacement of a protocol at one level should not require the modification of protocols of other levels. Tn addition, the protocol o~f any one level is trars- parent to the protocols of higher 1evels, i.e., does not intxoduce distortions into their work. '1'he protocols of higher levels (4 to 7) characterize the procedures ~ox direct (through ttie data transmission network) interaction of user systems with one an- other. Therefore, they are often called through ("end/end") protocols. Protocols of lower levels (1 to 3) describe the interaction of neighboring systems of any type. They, in addition, determine the standards for access to the data transmis- sion network. _ At the present time users widely employ programs of YeS [Unified Series] computers (SRV, OKA, KAMA, KROS, POISK, etc.). Therefore, at the first stage of development _ in computing networks it is advisable to.use at higher levels the standards and _ pr.ocedures of YeS computer tree-type teleprocessing. At later stages, of course, special virtual protocols, for terminals, filea, assignments, etc., wi11 appear in networks. The protocols of the physical, channel and network levels are described by CCITT Recommendation X.25. In keeping with it a data transmission network makes possible thc transmission of packets, and user systems and interface modules, before trans- mission, the breakdown of inessages i.nto packets and, a�ter transmission, t~e assem- bly of paclcets into messages. 2. Physical Structure Each system considered in table 2 is implemented in a complex consisting o~ one or more computers (large, medium, sma11 or micro), and each physical connection is accomplished by means of a data transmission channel. The ,functions of the inter- face module are per~ormed by an inter~ace converter which is constructed on the basi_s o� one or more mini- or microcomputers. The input/output module is imple- ment~d by means o~ a terminal. As a result, khe logical structure is converted into the physical s~ructure of the computing network, an example o~ which is illustrated in fig 6. Since r~ot only one but several syseems can be implemented in a complex, the number of types of com- plexes is considerably greater than the numbex o~ kinds o~ systems. Therefure, 56 FOR OFFdCIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFiCIAL USE ONLY ~ in a computin~ netw~~rk are used not on7.y' wQxking, texmina~,, comtpunication, and administrative, but terminal-woxking,, co~qnaunication- user, etc., co~qplexes. Com~unication compiexas (together with the inter~ace con- verters of any of them) ~orm packet switching centers. ~ ~ 6 A6oti:~e~ntx~p ~ - _ _ 5 tucmr~� ~ ~ _ 2_~ 1 - --7~ - ~ I X. ~S V,\ ~ _ j` ~ 3) a4) ~ 5)` , ~ - ~Yonn~HUKauuoN~na 3} 6 7 6~ cutme~n ~ 1) i , r a~ 4 ~ bM ~y 2 ~ ~ ~ p ~ b ~ - - ~ '~.1, ~ F b.. - - - - c~ ~q~ ~ - - - 1-- ' eR 3 I 67 RCMN~NUA"Q([UGHNQp I _ _ _ _Ct;C/1'~[NQ ' - ~ ~ ' a - -xzs ; ~ ~ ~ v.. g ~ ~ , ? ~ - - - - * il6oHeNmarnv - - - - - - - 6 CuCmer+4 7 Figure 5. Hierarchy o~ Protocols for Basic Conrrol o~ a Computing Network - Key: 1. protocols: applied (7); representation (6); network (5); transport (4) [Key continued or. following page] - 57 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY 2. User system 5~ Netwprk (3) 3. Physical (1) 6. Commun3cation sYstem 4. Channe~ (2) d~rruuumuw~uo cemb 1~ ~ Ctme 3BM 2, i ~ . ~ ~ ~ Q~ ~ ~ (eme nepccb~ ab~ar S~ ~ 3~ - / ~ ~ ~ 'c� I - . ~ ~ ~ % . I~ ~ ~ / i i 6 / / ~ ~ ' / `y~~ 1 ~ / ~ 4~ . , - ' ` _ _ _ - % ~ A6oNC,amcMUU nyN,rm Figure 6. Physical Structure of a Computing Network (For designations of 1 to 7 c~. text.) - Conventional symbols: ~ Working complex consisting o~ several computers ~.7 Single-computer working complex Q~+- Working complex with interface converter - ~ Terminal complex with terminals ~ Administrative complex ~ gacket-switching center (without inter,face converter and with interface converter) ~j Terminal complex with terminals and interface con- ~ verter - Data transmission channel with standard inte~c~ace ~ point Terminal [Key on following page) 58 - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY Key: Computing network 3.~ Aata transmission ne~work = 2. Computer network 4. User station 2.1. DaCa Txans~qission Networks Each data transmission network consists (cf. ~ig 6) o~ packet-sw3,tching centers and data transmission channels connecting these centers with one anothex as we11 as with user complexes or inter~ace converters. Large data transmission networks (cf. fig 7) can be mulfiilevel. In these networks the higher the level, the larger the capacity of the packet-switching cenrer and the higher the speed o~ data transmission channnels. The number of~levels o~ a data transtaission neework and its number o~ centers and channels are determined by economics� the topology and reliability o~ the network and by its speed characteristics. Points of the interface of packet-switching centers with user complexes and inter- face converters form the points of entry into the data transmission network. At these points are received or transmitted packets to be transmitted throughout the ~entire network without disassembly (or assembly). Packets are governed,by CCITT - Recommendation X.25. The entire data transmission network must be tran~sparent, i.e., must transmit packets the data in which can be coded by any method. _ ~ ~--~y~ , ~ i, . / - ~ ~ ~ Figure 7. Two-Level Data Transmiss:ion Network _ I [Continued on following page] 59 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 ruK ur'r'tLIAL USE ONLY Conventional syiqbols: . I,ax~e~capacity trunk line packet- switchiAg center ~ Zone packet-switching center A packet-switching center ~or making,possible high data processfng speeds is con- structed on the basis of mini- and microcotaputer~ and is a multi,processor center. Only for very small data tran~mission networks or for the lower levels o~ large hierarchical networks in which this is permitted by stipulations for reliability can a packet-switching center be a single-processor one. The entire processing of data in a packet-switching center is accomplished by means �of a working storage alone. The external storage (magnetic disks) is used only for loading the software and for gathering statistical data. In addition to packet switching, a switching center can provide for the performance of interface functions relating to the connection of non-atandard user complexes or terminals. For this purpose interface converters constructed on the bssis of mini- or microcomputers are included in the center. A data transmission network, representing a packet-switching network, can also provide for message switching (for compu~ers and teLegraph equipment). For this purpose, in the inputs and outputs of the network (centers b and d in fig 6) are added interface subsystems which perform the disassembly and assembly of messages. The assembly/disassembly of inessages is not carried ~ut at intermediate centers (a and c in fig b). 2.2. Working Complexes Working complexes make it possible to offer the most varied cnmputing and data processing resources to a wide range of users. Computers which have been created - specially for the network are used most effectively in them. The Ye5 and SM com- puters already produced were not designed for networks. However, by taking special measures the successful implementation of working systems can be provided for in these computers. Three methods exist for putting together a working complex constructed with one or more YeS or SM computers. The first af these consists in the fact that user com- plex 4(cf. fig 6) is connected via a standard MPD [data transmission multiplexer] or PTD [remote processing station] to the inter,face converter o~ comanunication com- plex b. This converter, implementtng the tnter~ace module, is in the guise of a standard user station f or the i~PD (PxD), and o~ a standard (,for the network) user , complex for communication complex b. The second u~ethod is J.ogical~,y equivalent to the ~irst. However, hexe the interface module is imp].emented not in the equipment linked with the communication complex, - but in the PTD, mini- or microcoipputer directly l~nked with user complex 2(cf. fig 6). 60 - FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 ~ FOR OFF1CiAL USE ONLY The third method o~ interfacing the working co~plex consiszs ~,n the ~act that special progxams making possi.ble the creation o~ the network method of access (cf. fig 2) are written ~or the xeS or SM computer~ Then, having added to network complex 1(cf. fig 6) a network adapter whtch responds to sta~daxd X~25, it is possible to connect the working complex direct~.y to communication complex a. _ In the computing network is also possible the combined utilization o~ a YeS computer - working with ordinary user stations and interacting with the network. This has been necessary primarily 3,n those situations when the XeS computer complex already had ordinary teleprocessing be~ore connection to the network. 2.3. Terminal Complexes - Termina.l complexes are created on the basis of YeS or SM computers or microcomputers. For cor,.nection to the network each computer must have a network adapter and hard- ware satisfying the protocols of the computing network. Network software is based on the utilization of standard operating systems and algorithmic languages of complexes. The chief software problem consists in organ- izing remote interaction between terminals used in the network with the resources of all working complexes present in the computer network. Three groups of terminal co:aplexes should be singled out: multiconsole, single- console and YeS computer user stations. Multiconsole terminal complexes are the most effective means of simultaneous con- tact between teams of users and the resources of a computing network (e.g., complex 3 in fig 6). The largest of them are constructed on the basis of YeS comput~rs. SM computers are used for creating the orhers. Single-console complexes are created on the basis of SM computers or microcomputers. - Each of them is designed for the work of a single user at a time. The YeS computer user station is a terminal complex (complex 6 in fig 6) including one or more terminals. It interfaces with the data transmission network via an interface converter created on the basis of a mini- or microcomputer. This type of terminal complex is the least effective, but it makes it possible to use YeS - computer character-user stations which the user already has. 2.4. Administrative Complexes _ An administrative complex implements an administrative sysCem forming the basis of the network contrql center. Each complex is coz~structed on the basis of one or - more mini- or microcomputers and it interacts with a11 or a group set aside for it _ of user complexes, communication com~lexes or inter~ace converters. If a single administrative complex is used in a computing network, then a11 the equipment in it is duplicated. for the purpose o~ ensuring high reliability. If there are severai administrative complexes in a network, then one of them is the - main one. When it goes out o~ order the functions o~ the main one are trans~erred to another administrative complex. 61 FOR OFF'ICIAL USE ONLY � : APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 h'UK UN'h'IC:IAL USE ONLY - 3. Topology of the Netwprk Let us discuss the principles o~ the creation o# a distributed comput~,ng network _ by using the example illustrated in ~ig 8. There are ~ive entities (eaterprises, institutions, management and organizati~ons) in which in~oxmation is developed or is used. Furthermore, the storage, processing and output of multi- purpose information, regardless o~ ita nature, must be accompl3,shed by m> 1 complexes. ~ s O~eewm 6 Afec.rm A . _ _ 2' . . ~ - 1~ g ~ 9 . I 4 \ ~ \ I ['ems w d~h Q3eam B ' ~ 6) - \ i ~ I . ~ 4) ' ~ ,e~ r f0 5 7 ~ ~ A?eam A ~ Figure 8. Topology of a Computing Network (cf. text for designations of _ 1 to lU) Conventional symbols: ~ User complex ~ Communication complex Key: 1. Entity A 4. Entity D 2. Entity B 5. ~ntity E 3. Entity C 6. Data txansmission network Depending on the 7.ocation of the points o~ the development and use of information, the information is divided (c~. ~ig.9) into l.oca7, and global.. I,ocal in~ormation is information which is developed and used at one and the same entity. But i~ in- formation is deve~oped in one and is used at least in one other entity, then it is global. In turn, two groups must also be distinguished among local information. Information which can load one ox more computers every day to a su~ficient 62 FOR OFFICIA~. USE ONLY i APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL U~E ONLY (economically petmissible) extent we wi11 ca11 steady. ~he xest o~ local in~orma- tion we wi11 place undex ~he head~ng o� peak. I Nxc~opMatiNa I , I. I . ~ ~ I - I JIoKaneHaa 2) I I3) I'no6anbt~as I ~ I - I Pae?~oMepxaa4)I I 'S)17NKOeaA I Figure 9. Classification of Information . Key: _ _ 1. Informa.tion 4. Steady 2. I.ocal 5. Peak - 3. Global - Steady local information must be processed (cf. fig 8) by computing complexes - - 1 to 7, installed at those entities where it is developed and used. Large, small or microcomputers can be installed here depending on the amount of this information. ~ As far as peak local and global. information is concerned, it must be processed in - computing complexes 8 to 10, the size, number and locations of which are determined _ by a feasibility study, taking into account the cost of equipment and communications channels, maintenance costs, reliability and efficiency requirements, etc. As a result a set of user complexes, 1 to 10, originates, located at the necessary _ geographic points and interacting with one another (cf. fig 8) via trans- mission network. 4. Local Networks _ - Local computing networks, which can opexate both independently and can represer~t users of large networks, have taken on ever greater importance. Here preferred development has been given to netwox'ks in which (cf. fig 10) a single packet- switching center is executed in the ~'orm o~ a monochannel to which network interface modu],es are connected. In turn, each ~qodule is connected to one or more user com- plexes. We will ca11 a network o~ this sort a monochannel network. Depending on zhe topology oX the computing network, the requixed data transmission - speed and the tra~fic, the monochannel is a twisted pair o~ telephone wires, a _ line, a coaxial cable, a light conductor, or the ether (radio). The length of a monochannel in small. networks uniting a group a~ instruments or microcomputers 63 FOR OFF[CIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 r~x ~rriuAL U~~; UNLY is not greater than one meter. ~n large 1oca1 networks containin$ dozens and ~ sometimes hundreds o~ computers the length of a monochanne~ can xeach several kilometers. 1) Mn - B iacyc~pcmDcHi~yw, - pcruo.+v.~eNya ~ubv 2~ ~av.wNya ctm~ Figure 10. Monochannel Computing Network Conventional symbols: ~ Working complex ~ Termina~ complex with terminals ~ Terminal complex with measuring or control equipment ~ Network inter~ace unit Y Internetwoxk converter Key: , . l. MP [internetwork convertez] 2. T~ State, regional or local network The network interface module is mast often created on the basis of a microcomputer ` and has the buf~ers required .for transmitting and receiving packet sequences. This makes it possible to accomplish the transmission o~ information through a monochannel at a rate di~ferent ~rom the data input/output rate in user complexes. 64 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2047/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY The method o~ accese to the monochannel is o~ great importance in an indepetident computing network. With a small nuu~ber o~' user comp].exes the netWOrk can haye - a supervisor untt which according to a selected cyc~e assigns comp7.exes the right to occupy the monochannel., However, with a considexab~~ number o.~ usez complexes this method becomes extremely ine~~icient. Thexe~ore, other ~ethods are employed i:t modern monochannel net~works. The most simpl2 method o~ access to a monochannel, called in [8] the "classical," is governed by the fo].lowing rules: 1) inter~ace unit enters in each packet a control sum making it possible to check the contents o~ this packet; 2) the packet is transmitted through the monochannel to the network interface unit addressed as soon as the neec~ for this arises; 3) the network inter~ace unit ad- dressed checks the packet re c~:ived and informs the transmitting unit i~ thexe are - errors in this packet; 4) with the existence of errors the packet is transmitted - again upon the expiration of a randomly selected time interval. Errors can appear in a packet because of the malfunctioning of equipment, noise, or, chiefly, the transmission into the monochannel of packets by several network - interface units at once. The repeate~ transmission af packets over various time ? intervals makes it possible tfl eliminate conflicts which arise. Losses of time for the repea~ed transmission of packets with heavy traffic can be rather great. There.fore, a number of inethods of improving the "classical" method exist. Considerable savings are produced by using the "slot" method, wttereby a packet can be transmitted into the monochannEl only upon the appearance of synchronization pulses from a central clock. The number of conflicts between transmitting units is reduced drastically if they also fulf ill the rule: "listen to the monochannel all the time and do not transmit if it is occupied." _ Experience has demonstrated that the creation of monochannel computing networks makes it possible to ensure the efficient processing of information ac an enterprise, in an institution, at a scientific institute and in an industrial association. Conclusion Thus, a computing network is a distributed multicomputer association designed for gathering, storing, transmitting, processing and outputting any kind of information relating to the control and functioning of various sectors of the na.tional economy. A network can involve the processing of a wide range o~ data or can be a special- purpose one d2signed ~or solv ing a narrow range o~ problems. The size of a co~puting network is determined by th~ probl,ems entzusted to it. Th~refore, on one hand, it can inc~.ude hundreds and even thousands of computers distributed ovex a large territory~embracing seyeral countries and, on the other, i t can be loca7. and have a to tal of a�ew s~qall, computers . The main tasks pex~ormed in a computing network are as .~oll,ows: the storage o~ _ data and the per~ormance o� data search operations; the mathema.tical modeling o,f 65 FOR OFFICIAL US~ ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY processes, machines, e~uipment, natura.7, pliez~p~ena and other entities; planning and the manageria,7, control o� variou~ e~emenxs .o~ the na~iona~, econo~qy; controll-~ ing production fiechnological processes; tha pex~orn4ance o~ booklteepi,ng a,nd statis- t~cal reporting ~or the ~unctioning o;C production unitsi cont~ol o~ scienti~ic research and technical, progress; the per`o:~~nce of computing (calculation) work; _ the distribution of stock; caxryin~ out e~ectronic librarp exchange work; contxoll- - ing the operation o~ savings banks; cpntrolling ciy~1 a~yiation; ticket sales and - reservations ~ox various kinds of transportation; a network o.f trade = centers; contxo~l,ing energy resouxces; etc. The users o~ a computing network, depending on the nature o~ the in~Eormation in it, are production workers, management apparatus coworkers, sc~enti~ic associates, and broad elements o~ the population. A11 the necessary methods o~ interaction be tween users and computing co~pl~xes are provided �or in a compu~ing network. Bibliography 1. Packet Switching Report. I~ogica Ltd, Sep tember 1978. 340 p. 2. Drozhzhinov, V.I., Ilyushin, A.I., Myamlin, A.N. and Shtarkman, V.S. "Principles of the Design of the SYeKOP Experimental Multiple-User Coniputer Network" in " kibernetiki. Problemy infvrmatsionnogo obmena v vychis- litel'nykh setyakh" [Problems in Cybernetics; Problems of Information Exchange - in Computing Networks], Moscow, 1979, pp 18-33. 3. riarchuk, G.I., Kuznetsov, Ye.P., Moskalev, O.V., Metlyayev, Yu.V. and Efros, L.B. "0 programme rabot po sozdaniyu vychislitel'nogo kompleksa (Tsentra) kollektivnogo pol'zovaniya v Novosibirskom nauchnom tsentre SO AN SSSR" - [Program of Work on Creation of a Multiple-User Computing.Complex (Center) at the USSR Academy of Sciences Siberian Division Novosibirsk Science Center], ~ Preprint No 130, Novosibirsk, 1978, pp 1-34. 4. Yakubaytis, E.A. "Arkhitektura vychislitel'nykh setey" [Architecture of Computing Networks), Moscow, Statistika, 1980, 278 pages. 5. Referencei4odel of Open Systems Interconnection ISO/T5, 97/SC, 16, N 227. . August 1979. 181 p. 6. CCITT Sixth Plenary Assembly. GeneVa, 1976. Orange Book. yol. 8. 2. Public Data Networks. Geneva, 1977. 2? 7 p. 7. P ublic Data Networks. Eurodata k'oundation. Lutyens House. k'insbury Circus, London, EC 2N( 71.Y. ].979. 233 p. 8. Cotton, I. Technol,ogies ~or local area cou{putex networks. - Computer Networks, 1980, N 4, p. 197-208. COPYRIGHT: zzdatel'stvo "Zinatne", ~~AVtp~atika i vychi;slitel'naya tekhnika", 1981 883I CSO : 1863/161 ~ 66 - FOR OFFICIAd. USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFiCIAL USE ONLY - ~ - UDC 681.324 DATA TRANSMISS~ON SYSTEM (SPD) OF THE 'EL'BRUS-1` MULTIPROCESSOR COMPUTING COMPLEX (r1VK) Riga AVTOMA,TIKA T VYCHTSLZTEL'NAYA TEKfIl~TIKA in Russian No 2, Mar-Apr 81 (manuscr~pt received 25 Apr 80) pp 15-30 [Article by V.S. Burtsev and V.I. Perekatov] [Text] It is obvious that the intense development of teleprocessing is now at the stage when the term "fourth-generation teleprocessing system" must come in use as applied to large-capacity computing conplexes and large comFuters. Experience gained in creating the fundamentals of the "E1'brss-1" MVK teleprocessing system and acquired as of the present time as the result of its development makes it possibl.e for us to formulate certain traits which distinguish the fourth-genera- tion system. 1. A teleprocessing system must represent a combination of general-purpose soft- ware and hardware making it possible to create practically any teleprocessing subsystem, including one whose key features were actually not drafted at the moment of the creation of these facilities. 2. The existence in the complex's structure of a problem-oriented program-con- trolled processor which solves a considerable portion of teleprocessing problems. - 3. The modular structure of the special-purpose processor. Modularity is under- ~ stood in the broader sense than simply the division of the processor into program- mable and nonprogrammable facilities. It is possible to speak of "vertical" and "horizontal" modularity. The former assumes the er.istence of a certain set of functional modules making it possible to develop a processor for solving one more class of probletus, e.g., a terminal module, a data transmission line ~odule, a transport level l:~ne module, a rapid-channel module, a module ~or controlling peri- pherals and the like. Horizontal modularity assumes the use in the structuxe of the processor o.f a variable number of modules of a single type (including modules o� othe.r types). Usually this number is determined by the number of communications linQs serviced and by the rate of the information flow. ~ 67 FOR OFFICIAd. USE ~ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 N'UR OFFICIAL USE ONLY - 4. Teleprocessing so~twaxe, including the so#twaxe o~ the speci,a].^purpose pro~ - cessor, is designed on the basis o~ ~zob~em-ox~,ented high-~.eyel la~guages, The distinctive #eatuxes cited do not pretend to co~pleteness~ They have been formulated chiezly in order to detqonstrate khat te].eprocessing systems have entered a qualitatively new stage of development and new trends require their own generalization. Teleprocessing systems characterized by the properties indicated above have opened up great possibilities. In particular, the question of the ability to use an MVK in a specific network architecture--clo~ed- or open-ended--has been eliminated. By using universal hardware and software which are inv~riant with respect to the principle of organization of the network, system personnel can adapt an MVK to a network of any type. The problems which arise with this are of an ordinary tech- nological nature and are to a great extent administrative. The domestic system characterized by the fundamentally new features of fourth- generation teleprocessing is the "E1'brus" MVK data transmission system. Able to serve as an example of a foreign fourth-generatian system is the DATACOM system implemented in the B6700-7700 series o~ the Burroughs firm. Fundamental Concepts As the res~ilt of an analysis o� the functions and operating features o~ an MVK SPD, two concepts have arisen which have been responsible for the beginning of a funda- mentally new development. The tirst of these--the concept of continuous adaptation--is based on the iollowing _ considerations. The list of remote users of such a large computing facility as the "E1'brus-1" '.rIVK includes equipment of the most varied classes--from teleprinters operating in terminal to computing complexes interacting in computer net- works. A cnaracteristic feature of large teleprocessing systems is their practical- ly contintious development--the elimination of obsolete components and the addition of new ones, the modificat~on of technical and logical attributes, the modification of interaction algorithms, and the like. In connection with this, the adaptation of. an MVK to remote users is regarded not as a one-time event, but as the way of life of the complex--a steady step-by-step process. The second r_oncept is the concept of primary processing. The fact is that in a number of applications an MVK is ~t~ the guis~ o~ a large in-line center for the processing of information arriving from communications lines. Fairly often the operations which thereby are performed on each received bit o� information (e.g., a character) are independent, i.e., do not require the use o~ data received earlier. A typical example is the screening of data, which is carried out according to some simp7.e criterion. In other cases processing is possible only with the presence of an group of received data (e. g. , ca~,cu],ation o� an arithmetic mean) , but it can be per~ormed gradually as each successive bit is received (the summing of data) and can be concl.uded immediately after reception of the last bit. Therefore, if the information load is considerable and the data r.ransmission rate is relatively s1ow, it is p~ssible to improve the eff iciency of an MVK by adding to its structure 68 FOR OFFICIAL USE UNLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL t?SE ONLY uncomplicated special,-purpose equipment (a eommunications processor) which pro- cesses the next bit o,~ in~ormation received regardJ,ess o~ whethex zhe ~o~lowing bit is received.* Zn certain cases zhis equipment can, as reception t~kes place, "without hurxying" per~or~ the ent-ire cq~puting task entrusted to the ~VK as applied to these data. In other words, primary processing ~hould ~ree the complicated equipment of the complex's central processor~ from the constant execution o� simple operations on small-format operands; this work con~idexably reduces the equipment utilization factor. On the basis o� these concepts decisive practical decisions were made and imple- mented: 1) to develop a programming language oriented toward the teleprocessing base level and universal within the framework of the problems o~ this level; and 2) to add to the MV'K's structure, as the equipYnent ex~cuting the base level's routine, a special-purpose data transmis~ion processor (PPD). The language has been given the name SFTRAN (for "setevoy translyator" [network translator]). It makes it possible for a system programmer or user to specify the entire network of remote MVK users, as well as the data and routines required for implementing interaction with them, e.g., the line discipline, the algorithms for exchange with a terminal, the variant of the LAR-2 procedure, a specific primary processing process, etc. To data t~ansmission processors (specifically to their adapters) are connected communications lines which connect the complex with remote users. Thus, the net- work is, as it were, broken down into several fragments, each o~ which is served by its own PPD (cf. fig 1). The SETRAN translator routine is executed in the complex's central system. The program describing the entire network is converted by the translator routine into several code files, each of which represents the operating system (OS) of one of the PPD's. A specific OS is designed to serve the network fragment of its "own" processor. Upon initiation it is loaded into _ the local working storage of the PPD and the processor becomes adapted to the fragment. Thus, for the personnel responsible for the creation of a specific teleprocessing subsystem for the "E1'brus" MVK the problem is reduced basically to an adequate J description of the network and of the basic teleprocessing processes in the SETRAN language. This made necessary the very care�ul coordination of the three key components: the language, the operating system procedures of the data trans- mission processor, and the PPD equipment. Actually, development took p.lace in integrated fashion, which made it possible to optimize a11 three components. *This processing px'inciple has been given the name "conveyer " processing. With regard to transmission the conveyez principle can be formulated in the ~ollowing manner: Each processed, i.e., prepared for transmission, bit o~ information is sent into the communications line regard~,ess of whether the ~oll.owing bit has been processed. - 69 FOR OFFICIA~, USE 4NLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-00850R040440020012-2 FOR OFFICIAL USE ONLY 2 ~ L(~Nmpanewo cucmerr MBK � .~Pyr - ~tir~om~op tempai,rs 3 ) 4) oc nnQO oc nnA~ a~ nRqN ceiryoNa nna o5) n~,q ~ . . . ,v _ 6) ~~'soy~- . noa noo'herm ~ 4 e~ ynb~eiweix 46owP.vrnoa MB/f a~ Figure 1. Principle of Adaptation of the "E1'brus" MVK to a Network oi: _ Remote Users Key : - 1. SETRAN program 5. PPDO 2. Central system of "E1'brus" 6. Switchable subnetwork MVK 7. Network of remote users 3. SETRAN translator routine 8. MVK - G. OS PPDO [operating system of data transmission processor No O] SETR~`.N: Concept of the Programming Language The SETRAN program consists of three sections which are described in succession: - the global description section, the network section and the algorithm section. It is best to begin a first acquaintance with the language with the network section. Network Section In this section is described the combi:nation o~ data processing modules participa- in the MVK's operation through communicat~ons lines. A so~tware-controlled remote unit intexacting with the MyK through a communica.tions line (in particular, a terminal) is an "external unit." ~ The second type o,� unit is the "line." We know, ~ox example, that in working with terminals connected to a mu7,tipoint line it is necessary to perform operations for 70 - FOR OFFICIAd, USE UNLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY - the line as a whole; an exautpl.e o~ these operations is poJ,lin$. Zn this sense a line can be singled out as a distinctive data processing unit coup~ed in a definite mannex with external units-~terminals. - 5ystem numbers o~ adapters, i.e., points .~or t,he connection of couunur?ications lines performing the role o.~ contacts between the network and the complex, are assigned �or units ox the "line" t}rpe. In the particu lar case of a single-point line the concept o~ an external unit and line unit often merge. Here the line unit can be used ~or so~tware control of the cluster formed by the remote unit and the communications line connecting it with the complex, Problems in basic teleprocessing sometimes reveal the need to control lines. ~or example, the in~ornation ,~lows of two counnunications lines duplicate one another for th e purpose o.f reliabiltty; therefore, only one of them, to be selected accord- - ing to a specif ic algorithm, must be admitted to the central system. A third type of unit--"internal units"--is added for problems of this kind. These can be re- garded as virtual data processing modules located in the MVK itsel~. Modules assigned in the network sectioii form structures called "trees In the structure of a tree one unit can be "nested" in another and, on the other hand, include severa 1 nested modules in itsel~. The implementation o~ SETRAN in the - "E1'brus" MVK permits six nesting Ievels. _ - As an example it is possible to take the section of a network illustrated in �ig 2a. It includes four communications lines connected with the adapters of a single PPD having the system numbers 0:1, 0:7, 1:4 and 1:7. Through the first line the = MVK in teracts with a remote unit, TERM. To the second line are connected three terminals, POLTERM. The third and fourth lines are connected with in~ormation sources which duplicate one another. Therefore, the necessity arises of adding to th e software a module which functions as a flow selector. In the network section this section can be represented by three trees whose struc- ture is shown in fig 2b. The first of these, a confl~ent tree, includes a single unit. The corresponding entry in the program has the following form: MODULE 4~ 0:1 TERM % Line mQdule OVER * MODULE 3~~ 0:2 TREKHTLINIYA Line unit. A~ter the procedure word [THREE-POINT LINE] MODULE is entered the number of nested modules--3. � . MODULE POL7,'ERMl . % E~ternal modu7.e OVER MODULE POLTERM2 - OVER � . [Continued on following page] 71 FOR OFFICIA~. USE ONY,Y APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 H'UK UhN'ICIAL USF. ONLY ~ " MODULE POLTE~3 flV~R _ OVER *MODULE 2 SELEKTOR [SELECTOR] % znternal module . . MODULE. 1:4 ISTOCHNIKI [SOURCEI] OVER ' r10DULE 1: 7 ZSTOCHNIK2 [ SOURCE2 ] OVER OVER 1> KKINTp[!.1/.N'J(~ Cu['T!n[ 2~ ~o4tccop ~rpv~lv~i c~Nma !"-~~~,..m~ 3 ) i . i' ni (~7 o~t ~ o: ~ 4~ ~j r r.�r~ I ~ 6 ) ~ 5~'�'" Cl ~O~m~/uy2 ~~nov.vur f Ntmo~.vur Y ~ /Ax~mtpn 3 a) 9 ~ Btiympe.wu! ~ . C~~~rmop y~ea ~ 10) 1~ n ~~e~m~uNUn y~,,,,,- ,,,,,uu - ---0-------- MCmovwu~ ! Ifrmovwu~ ? . /Ja.mtan 3 ` BN~u~rre ~�a�e~"'7 11~ y~..a AWmepn f � b) Figure 2. ~'rinciple of Representation o~ a Network of Remote Users in the SETR~IN k~rogram: a~-real netwcrk; b--structure of m~dules in network section [Key on following pageJ 72 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 ~ FOR OFFI~CIAL USE ONLY Key: 1. To central system 7. ISTOCtLi.iKl [SOU~iCE].] 2. Data transmission processor 8~ TREKIiTI~zNIX~, [THR~E^PQINT LINE] 3. Selector 9. Znternal modu].es 4. Adapters 10. I~~,ne modules 5. TERM 11. External modules 6. POLTERMI Here the dotted line replaces "contents of mvdule.�' The ~act is that each module is assigned specific progra~n components. ~or examp~.e, for the TREKHTLINIYA module they could have the form: * MODULE 3 4~ 0:7 TREKHTLINIYA MEMORY PArII~IN [MEMORY LINE] (POVTORY [REPEATS] : = 5) REGISTERS R$G QUEUE OCHLTN [QUEUE LINE] PROGRAM DISPETCHER [SUPERVISOR] [OCHLTN, 1&&255]; OPROS [POLLING]; VYBORKA [ACCESS] MODULE POLTERMI MEMQRY PAMTERM [MEMORY TERMINAL] MESSAGE ZAG [HEADING] (SB) QUEUE OCHTERM [QUEUE TERMINAL) PROGRAM WOD [INPUT]; VYVOD [OUTPUT] OVER MODULE POLTERM2 KOP [COPY] POLTERMI OVER MODULE FOLTERM3 KOP [COPY] POLTERMI OVER OVER This entry indicates that to the TREKHTLINIYA module are assigned the following: 1) its own memory, formed according to the PAMLIN mask determined in the global description section; 2) its own registers, formed according to the REG mask, also determined in the global description section; 3) the OCHLIN queue; 4) DISPETCHER, OPROS and VYBORKA routines whose texts are assigned in the algorithm section. These routines, of course, can be assigned to other modules; nevertheless, for the TREKHTLINIYA module they will be executed only in its context, i.e., will use variables Prom its own memory (registers) or its own queue; they are capable of activating another routine assigned by name only within the liznits of their own module. The context o~ any ~OLTERM module is broadex--in addition to the contents of its _ own module it includes the contents of the TREKHTLIN~YA module; generally the con- text includes the contents of a11 modules placed higher with regard to a branch of the tree. The contents o.~ POLTERM modules include also a message (SB) with a heading which is ~ormed according to the ZAG mask assigned in the global descrip- tion section. 73 - F~R OFFICIA~, USE UNLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 hUK Uh'h7~;lAL USE ONLY In the description o~ the POLTEiZM2 and POLTERM3 taodu7,es it is indicated that their contents are copied ~rom the POI,TER~l1 module. Thi,s abbreyiate~ the entxy. In the example given the contents o~ a module inc~.ude one queue and one meseage, but SETRAN in principle makes it poseible to asstgn to a module several queues or messages. Tt is possible to employ~unidimensional arrays of queues and messages; these are especially convenient in organizing buf~er networks. Unneeded components in the contents of a modu7,e can be absent. For example, it is possible to assign an internal module containing only a memory, or only queues, or a memory and queues. It wi11 represent a conrmon memory and common queues for the routines of nested line modules. Queues and routines (~or the purpose of activation) localized in nested modules also have access to routines executed in the context of a specific module, but for this purpose special facilities of the algorithm section are employed. It should be mentioned that routines run in the context of a line module and of any nested external module can be executed by receiving and transmission operators in conformity with the adapter assigned in the line module. Several adapters may be assigned in this module; accordingly, these routines can work with several communi- cations lines. _ Let the DISPETCHER [SUPERVISORY] program perform the decision functions in controll- ing a three-point line. This program must be activated if the central system orders some kind of operation (input or output) for one or more POLTERM terminals. These orders in the f orm of inessage units are attached to OCHTERM queues belonging to external modules. In addition, a message unit oriented directly to the line can arrive, such as for establishing a connection, or for disconnection or ~or performing a test. Messages o� this kind are included in the OCHLIN queue. A11 conditions for activation relative to the state of queues are assigned for the DISPETCHER program in square brackets. They can be understood as follows: If a module's routine is not executed and a message is referred to a vacant OCHLIN queue or to any vacant queue of any module located on the first nesting level, then the execution of this routine must begin.* - - The language provides for various methods o~ assigning individual activating queues and groups o.f them. The second method of activation assumes unconditional ~unning of the module's routine immediately af ter initialization o~ the PPD. By means of the third method one routine is started by a statement which is executed by another. For example, the WOD [INPUT] routine can be started from *In the 1&&255 construction the 1 digit indicates "first nesting ].evel" and the 255 digit, "any queue o~ the module." 74 FOR OFFICIAL USE UNLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFF~~^[AL USH: ONLY the OPROS [POLLZNG] routine, and the y'YvOD [OU'~~U~] routine �zom the ~Y~BORKA [ACCESS] routine. Thus, a structure corresponding to a multipoint line is assigned in the netwo rk section. With the presence o~ a task oxie~ted whplly to the 1.ine, a routine is started in the module which in the programmer~~ concept is associated with the co~?unications line. Having established logical connection with a specific termi-� nal, this routine starts the required routine in the module wnich is associated with the terminal, and the latter routine pex�oxms dtrect exchange. Everything said above must be regarded only as an illustration of several concepts and possibilities o~ the language. F~r exau~ple, the presence o~ a memory in POLTERM modules is ~usti~ied only in those cases when in a given exchange session the results of a preceding session are utilized in some way. I� this is not so, then the routines of each external module can operate with the data of a single common memory made available to the terminals by turns. This is the memory of the TREKHTLINIYA module. Similarly, a message in which cl-iaracters received from each terminal are sto red _ can be described in the TREKHTLINIYA module. And, of course, it is possible to describe the entire structure in the form of a single line module, whereby switching between terminals is accomplished by indexing structures be~.onging to this module. It is necessary to pay attention to one important factor. Having assigned the tree of the TREKHTLINIYA module in the network section and considering specif ic routines of the algorithm section in the context of modules of this tree, the programmer can completely separate himself from his surroundings (in this ca se the TERM and SELEKTOR modules) and assume that the entire task for the TREKHTLINIYA tr~e is performed by a certain virtual processor which operates with components which are described in the contents of modules. It attends only to communications lines connected to line module adapters. This principle, certainly convenient in solving problems at the basic teleprocessing level, was used for the firs t time in the NDL B6700 language. The description of the network in SETRAN also provides for the possibility of dynamic reconfiguration--the working replacement of the contents of modules and the switching o~ some fragment of the network for servicing by a standby PPD. Global Descriptioii Section "Masks" are de~ined in the global description section. A memory mask is ass igned in the form of a sequence of descriptions of simple memory variables and of uni- dimensional arrays o~ memory variables. On the basis o~ this description the language's trans],ator routine creates a cextain representation (mask) of the region of th~ meu~ory in which zepresentations o~ the variables to be describ ed have been packaged. Character locations of the mask are initialized in keeping with the descrintion of variables. ~f, then, in the network section the programmer assigns a memory mask identifier to certain modules, then the translatox routine wi11 actually isolate for each of them the physical region of the local working 75 FOR OFFICIA~, USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 N'UN UN'N'I~IAL USE ONI.Y storage of the P~D foz'nled accoxdin~ to the ass~,gned Iqask. A11 ~ axiab~,es whose representations aze local.ized in ~hi,s reg~oTt wi1,1 a~pea~ ~,n the modul,e's context. The programmer can in~luence packaging hy the pxecise ~ocation o~ a simple variable in the ~ask. Dixe~t addxess~;np ~s necessaxy ~ox the data trans- - mission processox operating s}rstem's intex~~ce and ~or routines of the central system. With specific principles floX the equipment~s structure, it makes it pos- sible to increase the ef~tc~ency of the data transmission processor on account of the simultaneous retrieval of several operands ~rom the memory. Relative addressing is a.1so poss~,ble, when the repxesentations o~ other variables are localized within the limits of a memory location storing a xepresentation of an encompassing variable. In fig 3, for example, is illustrated how the translator routine performs packaging on the basis of the descrip tion given. The programmer uses variables with the identifiers INFORMATSI'iA [INFORMATTON], TIPTNFORMATSII [TXPE OF INFORMATION], NOMABONENTA [NUMBER OF USER] and NOMPEREDACHI [NUMBER OF TRe1NSMISSION) in forming a heading, but when the need arises of traz?smitting into the cotmnunications line - a structure already ~ormed, he operates with the encompassing variable ZAGOLOVOK [HEADING]. �...~OIMOCOM ~B~Q~i MMIpOP`ip{(1/O w~~J ~ JOtM040R ~JJ~ 1 munu.~c{o~uu W~Q1 / wso~oAcur /61~ ~MV6o.w~ainv Q11,Q3/~otoeoBorl~J: ) ..o.,.,eP~"~ 9~ Q2 / ~oro.aea /11, . . . 3o~e.oeo. 2) ~ � 7 6 3 a 3 ? f p 3~� ~ ~ Nornrp[~abuu _ Ho~,v6o~rwmc (F~ ~~�~~~oP,.o~~u 5 ) N.~rOpMpuun ~ 6) - Figure 3. Principle o~ Combining Representations of Variables Key: 1. ...ZAGOLOVOK F8 D1; 3. NQM~k~~p~CHZ INk'ORMATSIYA F1 D1 / 4~ NQM~BQNEN~A ZAGOLOVOK [ 7 J; TIPIN�~ORMA- 5. ~IPTNk'OR~IATSzI TS I I F'1 D2 / ZAGOLOVOK [ 6]; 6. IN~'ORMATSIYA NOMABONENTA F1 D3 / ZAGOLQVOK [4); NUMPEREDACHZ k'1 A2 / ZAC~LOyOK [ 1 ] ; . . . 2. ZAGOI,OvOK The description can also no t include addressing o~ a variable; in this case the location of the repxesentation of the variable in the memory is determined by the translator routine. 76 FOR OFF[CIAd. USE UNLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY - A special type of simple u~emoxy ~rar~able is the sem~phoxe. Zt is highly pzobable that a variable o~ one and the same mask m~st be ~nitia~.~,zed di~~e~ently in the context of vax~ous modules. Let, for exa~rip~e, the u~axi~qum pe~cpqissib~e number of repetitions wi~th the unsuccessful transtuisSi,on o~ a po~~ing (or accessing) sequence be assigned by a variable with a Ppv~ORY [R~~F,A~S] ~,ndentifier from the ~AMLIN mask. Tn view o~ the vaxying quality o~ com~nunfcati,ons lines, the number o~ re- peats must be selected tndividually ~or each o~ them. This "tuning" o~ one or more variables of a mask as applfed to a module can be carxied out in the netwozk sec- tion. Above in describing the TREKIiTL~NTX'A tuodule we wrote - . . . MEMORY PAML TN (PQVTORY` : = 5 ) . . . This can be understood as ~ollows: the isolation for the module of a memory formed according to the ~AMI,IN mask, but with this the variable with the PpVTORY indenti- fier must be initf.alized not by the value assigned for it in describing the mask, but by the value 5. A second type of mask is "message masks." A message is a region of the memory - which is used fo r setting up interaction between routines run in modules of a single tree, as well as for interaction between any of them and routines of the central system. Having formed a message, one of the interacting routines transmits it to _ another; as a rule, this is done through a queue playing the role af a buf�er be- tween asynchronous processes. A message usually includes a heading and a text section. In the typical case of - a message unit for readout (from the central system) the text portion contains the text to be transmitted an3 the heading, certain parameters of the transmission pro- cedure; in the case of a message-result concering read-in (to the central system) the text portion contains the text received from the line and tt-~e heading represents the hard-copy log of the communications section. Messages containing only a heading are often used. _ For the SETRAN programmer the "message" is the message's identif ier. Actually, the message identifier equals the message descriptor--a location of the local working storage with a fixed address, which contains a reference to the message itself and - some information regarding it. Thus, under the same name the programmer can process an entire flow of inessages keyed to the descriptor at various times. In base level problems, as a rule, it is necessary to form or analyze a message heading. The programmer does this by using "message variables," whose representa- tions are localized in the heading of the message keyed to the descriptor at a - given moment. The SETRAN program can process messages with various heading struc- tures; the only limitation consists in the fact that messages with an identical - structure are to "pass" through a given descriptor. The message mask, which includes descriptions o~ message vaxiables, assigns the structure of the heading. In the network section each message identifier is assi~gned a message mask identifier (e.g., above ~or the POLTER~l module we wrote ...MESSAGE ZAG (SB)... , whez�e ZAG is the mask identi~ier and SB is the message identifier) . - 77 FOR OFFICIAd, USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 - - FOR OFFIC[AL USE ONLY Thus , the translator raut3,ne "knows" how to gafn acc~S~s to the head~ng (yia the descriptor) and how the xegion o~ the ~qe~ory~ ~ccup~ed by~ ~he heading is structured (from the description o~ the ~es.sage mask); th~,s ~ake~ ~C poss~b~e to pexkorm operations with components o~ the heading descxibed a~ tqessage Yax~ables, In communications processors is used nonprogra~nable line equipment which links the computer with communfcatfons lines.* zt is at the lowest ~unction- al 1eve1 of data transm~ssfon: ~t ~,nteracts dixect7,y w3,th cowqunications terminal equipment (in particular, with utodetqs), ffi the requixed duration of unit intervals, it gath~rs bits into characters in reception, and, on the other hand, breaks up characters in tran~mission, it eliminates (adds) start and stop pulses, and the like. Usually its mode of operation with a given communications line (adap ter) is determined by the sta~e of several registers accessible to a pro- grammable com~uter. The software facilities of these registers are assigned in a"register mask" in the f orm of a sequence of descriptions of "register variables." The r.egister mask identifier is indicated in the contents of the lin2 modulP after the procedural word REGISTRY [REGISTERS]. By operating with register variables, any routine of the line's tree can control the state of line equipment as applied to "its own" adapter. ~ In the register mask are also assigned virtual registers which lack a hardware _ equivalent; their functions are implemented by the computer through interpretation. - ' In other words, r~gister variables are used in individual cases as a means of "soldering" in~o the software those functions which were not added to the line equipment because of a of opportunity or simply on account of shortsighted- - ness. It must be mentioned that in addition to memory, message and register variables a"special variable" has been defined in the language. It is designated by the procedural word SIMVOL [CHARACTER] and represents the sof~ware ~acility for the location through which exchange between th~e computer and line equipment talces place . = Each adapter is assigned its own special variable, to which is imp.licity assigned - the representation of characters received through the adapter in execution (as _ applied to the adapter) of the reception sfiatement; thus they become program ` acc~ssible. In execution o~ the transmission statement the program-prepared re- presentation of the special variable is implicity sent through the adapter as the - - ne~t character to be transmitted; this operation can be performed more than once in a single statement. The constant mask contains descx~,ptions o~ simpl.e constants and o~ un~.ditqensional J arrays of constants. Zn the network section each ~~D ~s assigned an identi~ier *This equipment is present also in the "~1,'bru~-1" I~IyK ~PD. It ~.s called a ~ group interface unit (GUS). 78 - FOR OFFdCIAL USE 4NLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY of a certain constant mask. As the result o~ the const~nt o~ zhis ~sk, ar,y routine of the module run in a giYen ~PA can be used~ The "mask" concept as applied to cor~stants is rather arbitraxy, ~ox variants are , possible in which a eonstant mask cannot be ~.denti~ied the st~ucture of a - ~ memory region or o~' registers. Fnr example, ia the ~irst. implementation of SETRAN - the representations of simple constants whose word length is not gxeater than four bytes remain in the translator routine's tables and during g~neration are repre- - sented directly in the instruction code. Howe~rer, Tn the typical case the constant _ mask includes descriptions of arrays an.d o~ long constants. Thei~ representations are packaged in the 1oca1 working storage; there~ore, the translator routine, in - passing a description o~ the constant mask, creates a characte~ pattern of the region of the memory occupied by constants. The regfon of the memory structured _ according to this mask is actually isolated in the data transmission processor in translation of the network section. ~ _ In data transmission problems simple constants are especially important for ~ _ designating long strings and characters ~ahich do not have proper graphics. - Constant arrays can be described in one of two forms tn SETRAN. According to the ~ first form all elements of the array are described in the form of a list. Thus it is convenient to adc~ to the program code translation tables whereby the translation , itself Grill be perfurmed by an ordinary assignment statement, e.g.: CHARACTER: = K.DKOI7[SIMVOL]; here K.DKOI7 is the constant array with the identifier DKOI7 from constant mask K. The array represents a table of translation DKOI to KOI7. In the case of the second form, instead of a lit~ral indication of elements of the array is written the equation by which they are computed. It obligatorily contains a variable "ir,put element" (VKh) for which a iange of variation is assigned. Ele- ments of the array are generared by the translator routine, substitutitig in the equation the next value of the input element lying within the limits of the range. - Thus, the program in a number of cases does without routine work relating to the computation of factors to be assigned analytically. For example, let it be required to f ind the remainder from d~=viding (modulo-two) by a genera*ing F~lyr.omial a certain eight-bit code linked on the right with 16 zeros. With the step-by-step method of cyclic anti-interference coding this must be done after the reception o~ each charactex with a newly xormed eight-bit code. But it is possible, having sacri~ied the memory, to form a table of remainders ,~or all pcssible eight-bit combinations. T~ the genezating polynomial is given in the - same mask in the form o~ a simple constant with an identi~ier o~ X16X12X5X0, then for the purpose o~ generating the table it is su~~icient to describe the ~ollowing array: 79 _ FOR OFFICIAd, USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 ' rv~c urritr~wL UJ~ UIVLY ~ . . . [k'ROM 0 'PQ 255~ OSTATKT [RFaMAINDERS~ ~'8D2 ~ yKh&4~40000~>i`~OpMOA ~1,6X12XSX0;..~ ~ The last component of the global d.escriptf,on section is descxiptf.ons o~ ltne eve~its. A"line event" is the software form o~ an interrupt tx'ansmitted by the - line equipment computer. ~t can correspond to an actually generafied interrupt or to some virtual interrupt created by interpretation bp a"soldexed in" module - of the operating system. ?.lgorithm Section - The texts o~ module routines are described in the algorithm section. The addition to the language of subroutines .and functions has been workQd out and this has not - caused fundamental difficult~es. However, in the first edition of the translator routine sl~br.outines and functi~ns were not implemented because of the great overall amount o~ work, i~ Included among the n~umber of a.lgorithm section facilities which can be utilized in writing module program are primarily universal constructions such as structures and situations, conditional and selection (CASE) statements, synchronization state- ments and somewhat developed Dyks~ra primitives. ~ Ope:rations of 10 priorities have been defined for expressions. In addition to whole-number arithmetic operations, a number of logic operations have been added, as w~ll as specific operations which are especially useful in data transmission - p roblems, such as a shift (in particular., a cyclic), inversion, linking, embedding, d ivision and modulo-two addition and fon.~ation of a parity-check sequence. I n computing expressions is employed the b~.sic statement of the language according - to ~hich each construction used as a primary simultaneously has a"representation" and a"valtie." The representation of a construction is the continuous set of b inary digits assigned to it: 0 and 1. The value of a construction is tne value of the number assigned to it. By a number is nleant a positive whole number or zero. This 3ualism, making it possible to apply any opei:ation (in particular, arithmetic and logical) to any construction, is quite fruitful in base level teleprocessing p roblems. It is possible to cite many examples. L,~t us choose the most simple. Let the charac ter received be stored in a memory location "belonging" to the ~ ingle-byte va riable P.SIMV~L (i.e., to a variable wiCh the identi~ier STMVOL [CHARACTERJ ~rom mask The character enters the cot~unications line in KOI7 c ode. It is required to check whether it is the control character ICB (KONETS BLOKA [END O.r BLOCK]), whose code is assigned by the single-byte simp7.e constan~ K.KB. z~ it is, then certain operations are to be per.~ormed. Otherwise, the c haracter is t o be translated into DKOZ code and other opt~rations sre to be per- f ormed. It wi11 look like this: 80 FOR OFFICIA~. USE ~ONLY ~ APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USF. ONLY IF S IM~10L EKV [ EQU I~E1Lk~ ] R. KB ~ THEN . . . . . . . . _ OR ELSE ~.SzMVOL:= K.KOI7D[~.STMVpI,] . . . . . . . . . . . . . . . . . . . . . . . . OVER; ~ In the condi.tion is employed the 1og~cal operation "equi~valence" which checks the identity uf the codes o~ both o~erands. Tt works with the repxesentation of the P.SIM~'OL variable. In the OR ELSE alternative the translation of codes takes place. The translation table is the K.KOI7D constant array. In indexing the ~.SINfVOL variable array we use its value. It makPS it possible to select the element of the axray corres- ponding to the same character in DKOT code. Another example. Let there be crganized on the basis of the byte variab7le P.SCHETCHIK [COIINTER] a modulo-16 counter which registers the number of the informa- tion block to be transmitted. This number must be transmitted to the user before the block in DKOI code. In computing the current value of the counter an arithmetic addition operation is _ performed, utilizing the value ~f the variable P.SCHETCHIK:=*+l; but conversion - into DKOI code is performed by means of logical addition of the variable with a string having the code 11110000:P.SCHETCHIK:=*OR �11110000� . Here the re- presentation of the variable if utilized. A number of operators and standard functions make it possible to arrange for working with queues and messages: to attach and detach a message, to transf er all messages from one queue to another, to obtain informat3.on on a message and queue, to move the indicator of the current byte in a message, to store the next received - character in a message byte or to select the next message byte for transmission, to interrogate a region of the memory for a message (descriptor), and to transmit a message to routines of the central system or to the frAe memory region bank. This software is not fundamentally new--it is included in one form or another in DC-ALGOL B6700, for example. In SETRAN some of 3.t has been considerably developed or modified; that which did not require modification was taken without changes. The language necessitated the addition of several specific statements and standard. functions. Included in their number can be standard functions making it possible to work with queues and routines of nesred.modules,.~r the standard function - making it possible to convert a DKOZ string into a binary number and vice-versa. - - SETRAN is oriented towazd the conveyex method o~ data processing, Therefore, ; a special position among speci~ic ~acilities o,~ the algorithm section is occupied ~ by the reception operator and transmission operator, By util~.aing them the programmer organizes work with cominunications lines, which is in the final account _ the key function o~ the SETRAN program and og the dsta transmission processor. l-_ Tr 1~1 81 : = FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY Rereption and transmission operators fairl,y c,oulp~.ic~ted; there~oze, within the scope of this article it is begt to demonst~ate,~'their function in a brief example. . TO NETZAGOLOVKA [NO HEADING], PER~~OL~NENI7(~ [OV~Rk'LOW], OSHCHETNOSTI [PARITX ERROR], OSHKSB [BLOCK CONTR4L CHARAC~~R ERROR], A'~DNE'VNORM$ [DATA T'RANSMISSTON EQUIPMENT NOT UP TO STANDARD] %*The reception and transmission operators are combined. I;. is possible to build structures on their basis. Here is given is description of situa- tions % RECEIVE [1] MESSAGE SB [FROM P.NACHSEGMENTA [BEGINNING OF SEGMENT] TO P.KONSEGMENTA [END OF SEGMENT]] %*Reception is carried out from the adapter having a xelatiye number of 1 in the list of all line module adapters. Then within the ,franiework of the same operator characters received will be stored in the message segment whose boundaries are assigned by the value of the variables P.NACHSEGMENTA and P.KONSEGMENTA. Thus, it is possible to apply simultan- eously to a singls message several reception operators f illing various segments % - (SBOYAPD [DATA PROCESSING EQUIPMENT MALFUNCTION], KTOTAM [WHO THERE], DETEKTOR- KACHFSTVA [QLTALITY DETECTOR], CHETNOST' [PARITY]) %*These are "foreseen line elements," i.e., interrupts which can arrive from the line equipment instead of the next received character. Entered receiving medium. % * * FOR P.E'iALONZAGOLOVKA [STANDARD HEADING] CYCLE � � � � - � � � % *The dotted line indicates omitted constructions. % * - . ...NETZAGOLOVKA! [NO HEADING] REPEAT; %*Operator for start-stop cycle with a heading. Its function is to compare the sequence of characters (headings) received from the communications line with a certain standard, which in our case is assigned by the repre- sentation of the P.ETALONZAGOLOVKA variable. Each step of the start-stop cycle begins with the reception of the next character. At the end of the cycle's medium the character rece~.ved (possibly transformed in the cycle's medium) is compared with the corresponding byte o~ the variable. ~ With a lack o.f agreement the sequence o~ operators speci~ied after the colon is executed; in our case the operator �or a structural change with the NETZAGOLOVKA situation completes the execution o~ the entire reception operator. I� the entire heading is received success~ully, then the start- stop cycle operator is ended naturally TO VES'BLOK [ENTIRE BLOCK] , * CYCLE 82 FOR OFFICIA~, USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY %*The simple start-stop cy~cle operatox which will receive ~he information bl.ock is begun. ~he structure is cTeated on ~ts basis; the VES~BI,QK situation is described % - . . . . . . . . . . . . . . . . . . . IF SIMVOL [CHARACTER] EKy [EQUIVALENT]~K.KB THEN VES'BLOK! [ENTIRE BLOCK] OVER; _ %*In case of the reception of a character equivalent to the constant K.KB, we leave the cycle according to the ~TES'BLOK situation . . . . . . . . . . . . . . . . STORE IPEREPOLNENIYE! [OVERk'LOW] %*The character received is stored (possibly processed in the medium of the . start-stop cycle) in the "current byte" of the SB segment. If the segment - is already filled, then a structural change is perf ormed im~ accordance with the PEREPOLNENIYE situation and the reception operator is concluded - REPEAT; DELAY; %*Start-stop delay operator. The routine stops before reception of the next character % IF CHARACTER P.KSB THEN OSHKSB! [BLOCK CONTROL CHAR~ICTER ERROR] OVER %*The character received is a block control character (KSB). It is compared with the P.KSB variable "in which" the KSB of the block received is generated. With lack of agreement a structural change in accordance with _ the OSHKSB situation takes place and execution of the reception operator is concluded; ~therwise, the reception operator is concluded naturally.% (SBOYAPD : APDNEVNORME!,[(DATA TRANSMISSION EQUIPMENT MALFUNCTION : DATA TRANSMIS- - SION EQUIPMENT NOT UP TO STANDARD!,] - %*The "reaction to line events" entry begins. With the origin of a SBOYAPD line event a structural change is performed according to the APDNEVNORME situation and the reception operator is concluded DETEKTORKACHESTVA [QUALITY DETECTOR]: . . . . . . . . . . , %*With the origin of a DETEKTORKACHESTVA line event a certain sequence of operators is executed and the program returns to the same point ~f the reception medium where it was before the origin of the line event, i.e., awaits the next character % CHETNOST' [PARITX]:� � � OSHCHETNOSTI! [PARITX ERRORI %*In case of an error in parity of the character zeceived a certain sequence - of operators is Pxecuted and xeception is concluded as the result of a structural change in keeping with the OSHCHETNOSTI situation VSEPM [RECEFTION OVER] % The reception medium and the entire reception o~erator are completed. _ 83 FOR OFF[CIAL USE ONLY A APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 r~K ~rr~~tAL US~: UNLY � - WITH NETZAGOLOVKEI: [NO HE~zNG] . . . ~ . . . . , % *,P~ttachment o~ operation PER~POLN~NIXE: [OV~RFLOW~. � . . . . . , in keeping with situations OSHCHETNOSTI; [~A1ZIT'Y ERRQRI � , � � . . . , conc7.uding execution of OSHKSB: [BLOCK CQNTROL CHARACTER ~RROR]. th~e reception operator - APDNEVNORME [DATA PROCESSZNG ~QUTPMENT NOT UP TO STANDAIZD] : . . . . . . . . . . . . - VSESIT [.�i.TrUATIONS OVER] The transnt'.ssion operator is practically symmetric; by employing the start-stop cycle and start-stop delay operators and other ~acilities o~ the algorithm section in its medium it is possible to describe the transmission o~ a complicated informa- tion structure. Data Transmission Processor In lceeping with the general principle of modularity maintained in the "E1'brus" _ MVK, each specif ic complex, depending on the number of communications lines served _ and the rate of the information flow, can contain up to 16 PPD's [data transmission processors]. The data transmission processors interact with the central system of the MVK on the basis of a common working storage. The PPD gains access to the ' central working storage through an input/output processor (PW) which controls all external inf ormation flows of the MVK. Up to four PPD's are connected to a single PW (cf. fig 4). 1) nne ~ nee ~ . , , nea ~ 0 11 7 3 I~S 6 7 I? A H IS _ i 3~ ---4~-- nn4 i2~ r-- ~~-i i I ~ U^Kmpo~eNnKt hk~cmNao ~ ~ ~ ~ En~urnumree or~rpaiuO~vv i... ~ uB ~'10/J ~ ~ (l(B) ~unume i o r~...~s ~roon) I I�~ 5~ u 6 I O ? ~/yC ~ !f'C ' O~ ~ ~ ~ I ~ Acbrmipe~ ~ ~ I LO _IS _ _ ` I ~ ~ _ --...J L_--------~ Figure 4. Structure o~ "E1'bxus" MVK Data Transmiasion Processor ~ Key: 1 � PW 5 . TsV 2. PPD 6. M0~ 3. Central computer (TsV) 7. GUS [group interface unit] 4. Local working storage (MOP) 8. Adapters 84 FOR OFFICIAd. USE ~ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-00850R040440020012-2 FOR OFFICIAL USE ONLY , The data transmission processor can be connected direct7,y to two ~W~s. At a given moment it works on1.y with one o~ them, but when a~vV g0es out of order the in�ormation flow o~ the data transmission processor can be switched immediately to a s tandby PW . The data transmission processor has a modular structure. It consists of a central computer (TsV) , a locai working storage (I~Q~') and group inter~ace units (GUS's) whose number is determined by the number o~ coaununicat ions lines served. Group Interface Units The key funcCions of a GUS are as follows: to r~ceive data and convert it into character form for subsequent processing by the SETRAN program being run in the central computer; in transmission its function is the opposite--to convert charac- ters prepared by the program to a form in which they are to be transmitted through the communications line and then tc ensure their entry into the line. The transmission of data is usually performed in a serial bit-by-bit manner; there- fore, the function of a GUS can be defined more concis ely: the assembly of charac- ters in reception and their disassembly in transmission. Assembly-disassembly algorithms are determinate; only parameters can be changed--the type of data trans- mission (synchronous/asynchronous), the rate (bits/s) a.nd the character format. Accordingly, a GUS is designed as a nonprogrammable un it in which the parameters of the "soldered in" algorithms are assigned by the central computer, i.e., by the SETRAN program working with register variables. The structure of a GUS includes 16 adapter positions. An adapter of any type can be installed at any of them. Designwise an adapter rep resents a single standard element of the "El'brus" MVK. At the present time adapte~,:s o,f three types are used: 1) a telegraph, for direct interfacing~with the lin~~ (an Sl-TG interface); 2) the S2-100 ~V24 series 100) interface adapter; and 3) a parallel interface adapter (S3 and S2-200). All of these perform only the functions of an electrical , interface with the line or communications terminal equipment. The Kh-25 adapter, in which some logic is provided, is under development. An adapter position is tuned to a specific mode by ins talling the appropriate _ adapter and by means of software tuning of control fie lds which assign exchange parameters. The serial synchronous mode makes possible operation wi th speeds up to 9600 bits/s. An increase in speed is possible, but the maximum permissible number o� adapter positions in a GUS is thereby reduced. Two character formats--a main and addi- tional--are worked with direct].y. 7.'he additional format is used for the remainder of a coded program in cases when its total length is no t a multiple of tfie word length o.f the r.~ai.n format. One of the variants of the synchronous mode is designed especially for the cycle phasing procedure. With a steady reception mode there is a variant with the automatic exclusion o~ synchronization characters and with the transmission mode a variant with the automatic generation of these cha,rac ters. The code of a 85 FOR OFFICIA~, USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY _ synchronization charac~ez can be any oneR~it is a.ssigned bY the $ET~AN pro- ~ gram. The serial asynchronous mode is implemented ~or the range o~ standaxd speeds from 50 bits/s to 4800 bits/s and for a11 start~stop ~or~ts used in practice. Itg distinctive feature is the automati.c (without the participat~on o~ the central computer) readout of reception and trans~ission time-outs between characters. The length of a time-out is assigned by~ the progra~. The generation of an OTBOY (BREAK) pulse represents a separate mode. Its duration is software controlled. Possible emergency and nonemergency situations which arise in reception and trans- mission result in the generation of special interrupts transmitted to the central computer. They are processed as line events in the program. To GUS's have been added special gear making it possible to implement a program- controlled interface with communications terminal equipment (e.g., a modem). By their means the control of interface circuits has been brought up to the level of the SETRAN program, in which the appropriate procedure, in particular, the establishment of a connection, is described. A single group interface unit can be connected directly to two central computers. In case one central computer goes out of order the information flow of the GUS can be switched to a standby by means of special reconfiguration routines. Central Computer The central computer is a special-purpose processor designed basically to process information flows according to the conveyer method, A main feature of this equipment is the fact that it was developed under the great influence of the conceptually combined software, i.e., was designed for its imple- mentation from the beginning. Here special attention was paid to character-by- character processing--the execution of standard operations which {~,evitably or often correspond to the processing of each received or transmitted character. The hardware-controlled change of context must be singled out here. The fact is ~ that operation according to the conveyer method implicit~~ assumes ~he recurrence of routines carrying out character.-by-cKaracter processing. It is made possible by the fact that ,for each communications line (adaptez) is created a data module making it possible to organize execution o~ the routine in the context of precisely this line. Let us call it a"line context module" (BHI,). A standard component of a BKL is the address of the instruction ~rou4 which it is necessary to begin exe- cution of the routine Frith the arrival o~ the next ~haracter. Ordinarily a11 BKL's are stored in a local working storage in the form o~ a regular , structure. The ef~iciency o~ the processor is impxoved markedly i;~ the BKL ~or the line served is trans.ferred to registers with a short access time during the processing o� a character. Accordingiy, after the completion of proc~ssing the _ BKL must be returned to the MOP from the rapid register zone. Tn implementations 86 FOR OFFICIA~. US~ ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02109: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY of the conveyer method known to the authoxs (e.g., in the AC~ B6700-7700) the transfer o~ BKL~s is performed by the opexatin~ sys~em, i.e.t is Sq~twaxe con- trolled. This takes a considerable amount of time and inevitab~,e oyerhead costs originate. - In the "E1'brus" MVK PPD a change of line context is made posaible by the hardware: The process of requesting and returning the BKL is per~ormed automatically and - coincides with processing o~ the next character. Thus, the central computer begins the operations with the next character without preparatory operations, immediately after completing processing of tt~e preceding. Sof tware requirements have, of course, been reflected in the instruction set. The four-byte instructions o� the central computer are oriented toward the execution of SETRAN operations on byte operands and operands representing a group of bytes. They make it possible to speed up access to program constructions localized in _ the local working storage and the storage o� the central system and, :tn particular, to organize working with queues and messages. By means of specific instructions a very simple mechanism for requesting procedures has been implemented, making it possible to structure not too badly the object code of the SETRAN program. Conditional transfer instructions are oriented toward the most typical software situations--the arrival from GUS's of interrupts of a specific type and code, a change in the state of f lags and the like. Advanced hardware control has been added to the data transmission processor, which practically eliminates the pos~ibility of malfunctioning or failure without their instantaneous indication. As the result of an emergency situation the hardware automatically causes special reaction routines representing part of the operating system of the PPB. The program reaction to malfunctions makes it possible in prin- ciple to maintain the survivability of the PPD right up to extreme emergency situ- ations. Conclusion _ Results achieved in the creation of the data transmission system representing the basis of the entire "El'brus" MVK teleprocess~:ng system have been described in fairly c~mplete form in this study. It, of course, will be subjected to changes in the process of gaining experience in using it. However, if allowance is made for the fact that the "E1'brus" MVK SPD [data transmission s~~stem] is in the overall course of the development o~ fourth-generation teleprocessing systems, it can be considered a fundamental basis ~or the creation of further studies in this area. - COPYRIGHT: Izdatel'stvo "Zinatne", "L~vtomatika i vychislitel'naya tekhnika", 1981 8831 CSO: 1863/161 87 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02149: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY ~ UDC 621.391:621.395:681.142 NEW BOOK DISCUSSES CONTROL SYSTEMS FOR COMMUNICATIONS NETWORKS Moscow SISTEMY UPRAVLENIYA SETYAMI in Russian 1980 (signed to press 21 Jun 80) p 138 [Table of contents of book "Control Systems for Networks", edited by Pro- fessor V. G. Lazarev, doctor of technical sciences, and N. Ya. Parsfienov, candidate of tecl`~nical sciences, USSR Academy of Sciences, Izdatel'stvo "Nauka", 1,650 copies, 140 Pages] [Text] Table of Contents pagP Parshenkov, N. Ya., "The Probablistic Game Method of Dynamic Conrol of Flows in Channel Switching Networks" 3 Umrikhin, Yu. D., "Some Theoretical and Applied Aspects of the Systems Approach to Synthesizing Adaptive Control of a Data Transmission and Distribution Network" 9 Tavuzhnyanskiy, G. D., "Control of Information Flows at Intersections of the Network with Packet Switching" 36 Kotov, A. V., "The Question of t+~~ Systems and Comprehensive Approaches in Telephony" . . . . . . . . . . . . . . . . . . . . 41 Koshelev, V. N., "Hybrid Switching Systems" . . . . . . . . . . . . 52 D'yachenko, V. F., and Chernyayev, V. G., "Constructing - a Message Queue at a Switching Center Using Uneven Sets" 62 Solov'yev, A. V., "One Way of Evaluating the Time Required to Service Calls in the Electranic Control Machine" i0 Gol'dshteyn, B. S., "Optimal Priority Service in the Software of the Electronic Automatic Telephone Exchange" . . . . . . . . . 73 Doniants, V. N., and Piyl', Ye. I., "Synthesis of Boolean Functions in Modified Matrix-Type Homogeneous Media with Repetition of Variables" . . . . . . . . . . . . . . . . . . 79 88 FOR OFFICIAL U5E ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY Page Yanshina, L. K., "Selecting the Number of Coding Variables in a Nonautonomous Microprogram Automatic Device" . . . . . . . . . 90 Melikyan, E. R., "Constructing and Selecting the Heirarchical ~ Structure of a Control System with a Minimum Number of Control or Information Links for a Given Productivity" 94 Turuta, Ye. N., "Analysis of the Reliability of the Conveyor- Type Multimicroprocessor Control System" . . . . . . . . . . . . . . 111 Yershov, V. A., "Average Accessibility and and Recurrent k Calculation of Losses in Switching Systems" . . . . . . . . . . . . 121 _ Kuusk, E. E., "Using the Computer To Determine the Intensity of Telephone Load and Losses" . . . . . . . . . . . . . . . . . . . 128 Kholin, A. V., "Physical Modeling of Communications Networks" 133 COPYRIGHT: Izd~.el'stvo "Nauka", 1980 11176 CSO: 1863/137 _ 89 FOR OFFICIA~. USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFI~CIAL USE ONLY UDC 621.391 HYBRID SWITCHING SYSTEMS Moscow SISTEMY UPRAVLENIYA SETYAMI in Russian 1980 (signed to press 21 Jun 80) pp 52-62 [Chapter by V. N. Koshelev of book "Control Systems for Networks", ~dited ~ by Professor V. G. Lazarev, doctor of technical sciences, and N. Ya. Parshenkov, candidate of technical sciences, USSR Academy of Sciences, Izdatel'stvo "Nauka", 1,650 copies, 140 pages] _ [Excerpts] Work began in the late 1960's and early 1970's to build hybrid switching systems. The need to build integrated communications networks and hybrid switching systems arises, first, from the fact that the develop- ment of technology and data processing shows a growing need for information - exchange among different kinds of users. In the second place, as studies [6, 9, and 10] show, integrated communications networks are more economical than other types. This economy is achieved by fuller use of expensive main channels. In conformity with the recommendations of the International Telegraph and Telephone Consultative Committee, data transmission equipment is classified by transmission speed into the following groupa: low-speed - 50, 100, and 200 bauds; medium-speed - 600 and 1,200 bauds (speeds of 1,800, 2,400, 3,600, and 4,d00 bauds are also permitted); high-speed - 19,600 and 48,000 bauds and higher. Information in a communic3tion network may also have different character- istics. There are short messagea transmitted in a mode close to real time, long messages in real time, and large arrays of data. Information trans- mission ma.y be Performed improved reliability or without, with . due regard for the priority of the user and information or without. The different characteristics of user terminals and information being trans- mitted require that communications networks have different characteristics and that the r~witching center have different functional features. Thus, the future communications network should be suitable for serving different groups of users, which will require that the network have spe- cific characteristics that differ from exiating networks. These include a broad range of data transmission speeds, the possibility of transmitting communications in a wide range of lengths, and flexible protocol that makes 90 ~ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY it p~ssible to work in real time or relative time scales among different types of users. Integrated communications networks are designed and built on the basis of already-existing networks. Combinations of switching procedures in inte- grated communicationa networks are e~ployed with due regard for integration of functions and joint use of the resources of switching pro- cedures. The problem of constructing a general-use integ~ated communications network - covers a number of questions: the architecture of the network, linking the network and the user, network control, consolidating lines, and constructing switching centers. The main question in constructing an integrated communications network, which in large part determines the solution to the other questions, is de- fining the degree of integration of the communications netk*orks. Figure 1 below shows different degrees of integration for channel and packet switch- ing networks. Q~~.~, nuydu id~v~u ~t'~t' ~ uHih tpane~eit J ~ ~ru,rau cs.rru. nub`u > Ca.~u ,C/! I ~t'/I K/I _ - (3 ~ nuyuL2~d.rr~ ~t'~C ~ ~K ~ ~ KK nuyau cssru i - K~ ~ n~~~~ ~e...,~~~ Figure 1. Key: (1) Switching Channel (KK); (2) Communications i.ines of Switching Channel; (3) Packet Switching (K1i) ; (4) Packet Switching Communications Lines; (5) Integrated Communications Lines; (6) Communications Lines. Figure la shows the divided structure of communications networks that exists at the pr~sent time. Each network here has its own set of users and com- munications lines. The co~unications networks do not have common equip- ment. The first degree of integration (Figure lb) offe;.s users the possibility of employing all the resources of the communications networks tfiat~ma~Ce up the integrated system. In this case the switching equipment at the switching 91 FOR OFFIC~AL USE O~1L: ~ APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 FOR OFFIiCIAL USE ONLY center and the communications lines renain separated for all switching methods. Users are connected to the co~unications network through one of the switching systems. In this case it is the channel switching system. The second degree of integration (Figure lc) has maximal merging of the func- tions and resources of the switching systems being combined. It is based on one system and minimizes the individual equipment of the other. For example, - if the integrated communications network is constructed of channel switching and packet switching systems based on the channel switching system, the indi- vidual equipment of the packet switching system is minimized. With this - de~ree of integration the connecting lines are shared, while the switching and control systems remain separate. Finally, the third degree of integration offers the possibility of access to the equipment of the communications network for users in all cate- gories (see Figure ld) with shared use of communications lines and merging the switching and control functions for all the switching systems being consolidated. This permits shared use of the entire memo.ry volume of the memory unit, the power of center processors, and the software system of the switching center. . Development of the switching center is a key question in building an inte- grated communications network. When selecting the structure of the switch- ing center it is necessary to consider the similarity and differences of the processes of establishing connections for different switching proce- dures. The similarity consists in the existence of three phases of processing a demand to establish a connection and transmit information: ~ the phase of establishing the connection; the information transmission phase; the signoff phase. The procedure for initiating the communications channel is also the same for all methods of switching. The difference is that with some switching methods (channel switching, and sometimes also packet switching) a real communications channel is established, while in _ other cases (packet switching and communications channels) the channel is virtual. This leads to a difference in the organization~of hardware and so�tware. The software includes additional procedures to or~anize the distribution of inemory to store packets and messages, as well as the pro- grams needed to process them. The hardware includes the additional ~ memory volume of the center memory units and the additional power output of the center processor. The additional power expenditure of the center processor is reduced by cons~ructing and separating the modules for processing and storing packets and messages from the center processor. This greatly increases the tlexibility of the hybrid switching system. Furthermore, the additional power expenditure of the center processor is reduced by constructing a multiprocessor str+~cture in the switching center [4J. The next important question of constructing the integrated communications system, which determines the choice of the switching center structure, is selecting the method of hybridization. Several techniques are known: separate receipt and processing of information transmitted by different switching techniques; joining the information of the data transmission - network and telephone network users into "envelopes" which are transmitted 92 _ ~ FOR OFFICIAL USE ONLY - APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007102/09: CIA-RDP82-00850R000400020012-2 I FOR OFFICIAL USE ONLY on the main line, and transmitting information with communications channels - and switching channels with transduction in the hyhrid switcfiing center (but for communications channels complete messages are transduced, whereas with switching channels it is particularly meaningful segments). At the present time hybrid switching sys tems for integrated communications networks are being developed and introduc ed intensively. Some of the most recently developed systems are the EDS and FDX systems which are being introduced in the national netwo:k for t ransmission of discrete informa- tion of West Germany (4, 11, 12]; the APB-20 and T-200 in Switzerland [8, 16]; the Unif ied Switching Center and DS-714 in England [7, 14]; the DDX-2 in Japan [13]; and, the M-3200 in the United States [5, 15]. Let us consider a few of them. Conclusion Our analysis of the experience of many large companies in the areas of - designing, developing, and introducing hybrid switching networks enables us to formulate the following basic requirements, whicfi must Fie the basis _ for construction of such networks: 1. Optimization of the process es of controlling blocks _ of rransmitted information; 2. The existence of general netraork protocol that defines the principles of control of the network and interaction between network centers and users; 3. Selection of optimal switching procedures depending on ~ the characteristics of the information, users, and con- dition of the network; 4. Insuring a high degree of re liability in switching centers; _ 5. Insuring the possibility of expanding the capacity of switching centers; 6. Ability to operate in hoth real and relative time - scales; 7. Developing software and a system of commands that maximally simplify programming and the entire cycle of software work in the process of information ex- change. COPYRIGHT: Izdatel'stvo "Nauka", 1980 11176 CSO: 1863/137 - 93 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPROVED FOR RELEASE: 2007/02/49: CIA-RDP82-44850R000400024412-2 FOR OFFICIAL USE ONLY UDC 681,32:681. A METHODOLOGY FOR THE SIMULATION MODELING OF THE COLLECTIVE-USE COMPUTER CENTER OF THE SIBERIAN DEPARTMENT OF THE USSR ACADEMY OF SCIENCES Riga AVTOMATIKA I VYCHISLITEL'NAYA TEKHNIKA in Russian No 1, Jan-Feb 81 pp 3-14 manuscript received 25 Dec 79 [Paper by Yu . I. Mitrof anov] [Text] A number of systems analysis problems came up during the design of the collective-use computer center of the Siberian Department of the USSR Academy of Sciences (VTsKP) [CUCC] being created in the Novosibirsk Scientific Center, the solution of which proved to be possible only by means of simulation modeling. In terms of algorithm structure and functioning, the CUCC is a typical network of computer complexes [1] and belongs to the class of large complex systems characterized by stochastic functioning. After studying the possible languaga tools for the simulation modeling of the CUCC, ALGOL-60 was chosen with the procedural expansion up to the level of a timewise modeling language for discrete systems (the set of simulation modeling procedures f or discrete systems - the KIMDS) [2, 3]. The reasons for such a selection are treated in section 1. The methodology for the simulation modeling' of discrete systems using the tools of the complex, its composition and structure whicti was realized in the set of simulation modeling procedures for discrete systems are treated in section 2. The organizational principles based on the set of simulation modeling procedure~ for discrete systems for the CUCC simula- tion models are treated in section 3. The composition and structure of two simulation models of the CUCC are discussed in section 4 as well as some results of studying it. 1. Language tooZs for simuZation modeZing. Simulation modeling as a technique for studying real and conceptual systems provides the possibility of constructing models which a theore~ical description of the systems being modeled with an level of detailing. It is specifically this fundamental property that in many cases determines simuiation modeling as the only possible study technique (especially for complex systems), despite its well kr~own drawbacks: the labor intensity of the development of simulation models, the complexity of debugging 94 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY them and the large amount of computer time required to perform experiments with models on a computer. The development of the means of automating simulation modeling an3 computer hardware is constantly reducing the relative weight of ttle deficiencies of this method as compared to its advantages, thereby providing for a constant expansion of the practical utilization of simulation modeling. The labor intensity of the development of simulation models and their debugging depends substantially on the programming languages used for the models, in particular, on. the existence of convenient and economical tools in these languages for representing the systems being modeled in models of the structural and functional properties. When deve'loping the m~ans of automation for simula- tion modeling, as a rule, the f act that the potential users of these tools are usually specialists in the systems being studied is taken into account, where these specialists do not have professional training in the general case either in programming or in the methodology of simulation modeling. For this reason, the modeling tools should basically allow for the reduction of the task of developing a simulation model to the convenient and natural depiction of objects and relationships in the system being modeled by the appropriate objects and relationships in the modeling system being employed, In this case, the method- ological principles for the construction of models are governed by the systems being used for the automation of the modeling. For example, discrete system modeling languages are being used at the present time which are oriented towards the depiction of events, processes, operations, etc. [4~. Universal algorithmic programming languages (AYaP) by virtue of their primary function neither contain the specific tools necessary for the represenCation of time relationships in the models, where such time relationships occur in the systems being modeled, nor embody a methodology for the construction of simulation models. At the same time, the broad capabilities of these languages as regards the representation of functioning algorithms for the elements of the systems being modeled have made it possible to use them as basic languages in the development of a number of specialized modeling languages (SYaM) (for example, SIM[TLA and SIMPL/1). At the present time, both universal algc~rithmic programming languages and special- ized programming languages are used in s~mulation modeling practice for discrete - systems (the advantages and drawbacks of their use for these purposes were treated in [4]). The reasons f or the use of an algorithmic programming language in simulation modeling are actually deeper than those usually cited (the lack of translators from the modeling languages to those accessible to a computer user, difficulties in operating modeling systems, the lack of high quality documenta- tion for users of modeling systems). In comparing algorithmic programming languages and specialized modeling languages from the viewpoint of their usage in simulation modeling, the followi~ig must be taken into account: 1) The ideas and content of the basic concepts of simulation modeling are rather simple, and understanding them does not as a rule cause any difficulties; 2) The existing modeling systems are oriented to one degree or another towards definite classes of modeled ssytems, something which generates differences of a methodological nature in the depiction of the modeled systems in the models; for this reason, in particular, recommendations for users regarding the choice of specific modeling systems prove to be useful [8]; 3) The striving of designers of 95 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFF(C1AL USE ONLY modeling languages to make them universal, on one hand makes it difficult to study these languages and their practical applications by users, who are not programming specialists, and on the other hand, reduces the efficiency of the , simulation models as programs; 4) In the development of simulation models for camplex s~stems, along with the complexity of the functioning algorithms for the _ system components, the weight of the tools made available by the modeling language ~ for the description of the algorithms increases with respect to the tools for the systemic organization of the models. Thus, the utilization of universal algorithmic programming languages (as a rule, with the creation of the requisite tools within the framework of these languages) in the case of simulation modeling of complicated discrete systems is natural at the present stage of development of specialized modeliug Ianguages, which is on the whole still not satisfactory. 2. The KIMDS 5imuZation ModeZing CompZex. In structural terms, the ser of simu- lation rr.odeling procedures for discrete systems is a camplex of procedures for ` various purposes [3]. An independent component of the complex is the master program, which organizes the operation of the simulation models as a whole. All _ of the procedures of the complex can be broken down into several sets: synchron- ization and control procedures; statistical data retrieval and processing pro- cedures during modeling; random quantity generation procedures; procedures for working with data f iles; and service procedures. More than 60 procedures for different f unctions are incorporated in the complex at the present time. The structure of the complex makes it possible, on one hand, to expand the composition of the camplex through the inclusion of the requisite _ procedures, and on the other hand, to utilize only the requis:ite procedures of the complex in the development of specif ic simulation models. Ttie States of Processes and the Control of Their Development. We shall consider the basic principles for the organization of simulation models of discrete systems using the example of a simulation model I for a system S. We call a structural unit of the simulation model a program process. Corresponding to each element Ei, i= 1, 2, n, of the system in the simulation model is _ a program process Pi, which is a description of the given element jn the ALGOL-60 programming ~anguage and formatted as a procedure. Consequently, n program processes Pi, which are descriptions of the elements Ei, are incorporated in the model I. Messages are represented in simulation models by requirements of various classes, while the message sections are represented by requirement sections. The require- ments in the models are represented by their numbers, and for this reason, the motion of requirements in the models is essentially the motion of the numbers representing them among the various units of the models. Setting the requirements in a section and selecting them from the sections in the models is accomplished by accessing the put and get procedures. We shall assume that all of the processes in the model are numbered, and the number of the process Pi is designated as vi. 9~ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFF[C1AL USE ONLY ~ ` We shall draw a distinction between the main and auxiliary possible states of the program process Pi and designate them as p~f~, s= 1 2, 10. The main states are the active and the passive states (p~l) and pt2~ respectively). The remaining states are auxiliary ones. When running a simulation model on a computer, each program process realizes a certain sequence of discrete events on the model time axis, which represents the corresponding element3ry process. The realization of the events is accomplished by program processes only at the - point in time where are present in the active state. If some condition must be met to realize the next event, then the process is in the passive state until this condition is met. We shall term the state p~3~ the stop state. It is a special case of the passive state, when the condition for the activation of the process is the direct action of another process which is in the active state. The introduction of the state p~~), 4, S, 10, was needed for auxiliary purposes related the operational algorithms of the modeling system as a whole, and in parcicular, to make an analysis to see that the events ~eing realized in the simulation models are noncontradictory. Synchronizati.on and control procedures, as well as operators f or accessing them . are used in the bodies of the processes to assure the representation of the _ parallel development of elementary processes in a model and provide for the synchronization of the development of the program processes as well as control their development. We shall define the content of the synchronization and control procedures, presupposing for the sake of definiteness that the operators f or accessing them ~ are used in the body of the process Pi: delay (p, ~y, w) delays the development - of the process Pi (shifts to the passive state) for � units of model time or until the condition p is met; hold w) delays the development of the process Pi for � units of model time; wait (p, w) delays the development of the process Pi until the condition p is met; halt (v w) stops the process at the number v(shifts to the stop state); leav(w) stops the process Pi; start(v, w) activates the process at the number v(shifts to the active state), which is in the stop state. The realization of the a,lgorithm of the process with the number v will be started with the first segment of the process algorithm; push(v, w) activates the process with the number v, which is in the stop state (the realization of the algorithm for the process with the number v will be started with the segment of the algorithm, aheald of which it was stopped); finish shuts down the modeling process, i.e., terminates the execution of the simulation model on tiie computer. Service Tables and the Master Program. The operation of the set of simu- lation modeling procedures for discrete systems anci the simulation model as a whole is ~ontrolled by the master program, which, just as the procedures for syn- chronization and control, uses information on the processes during operation which is contained ii: tables of process states, time markers, ~eturn poiiits, prototypes and copies. In these tables, the numbers of the elements which contain the infor- mation pertaining the processes correspand to the numbers vi of the processes Pi. The codes of the process states p~s), s= 1, 2, , 10, are the values of the ~ elements of the table of process states. 97 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY We shall use A to designate the current value of the model time, and T for the _ point in the model time at which the next event should be realized in the process P(we shall call this mament the time marker of the procesa P). We shall use the character to designate a positi~~~ real number which is the representation of ~ in the digital computer, on which the modeling ie gerformed ($n experiment with ' a simulation mo%el). For processes which are in an active state, T~ 9; in a passive state r> A(in particular, we can have T m[~?, and in the stop state, t= The process P can shift from the active state to the passive one only as - as a result of the execution of the operator of one of the synchronization proce- dures: delay, hold or wait. With the execution of the f irst two operators, the point in time T for the process P is camputed from the formula T= A+~; with the execution of the operator for the wait procedure, T takes on the value of The transition of a process P with a number v to the stop state can occur either from the active state in accordance with the operator for the leav procedure, or from the active and passive states with the operator for the halt procedure, carried out by another process. With the transition of the process P to the stop state, T assumes the value of The values of the elements ~f the time marker table are the points in time T for all of the simulation model processes; these values fall in a range of [6, It stands to reason that the values of T for each process change in the course of running the routines for the simulation model on _ the computer. During the execution of the operators of the synchronization and control procedures, used in the body of the process P, the control is transf erred to the master program. To provide for the return of the control to the process P, the master program should have data on the "point of return to the process" or the number of the algorithm segment which should be executed as the next one and which directly follows the operator. The point of return to the process is governed by the parameter w in the synchronization and control operators. The value of this parameter is entered by the synchronization and control procedures in the table of return points, i.e., the value w is ass~gned to the table element corresponding to the process P. ~ We shall call two process Pi and P~, i. 3= 1, 2, , n, i~ j, equivalent if they differ only in the values of the internal parameters. We shall designate the number af equivalent processes of the k-th type as ek, k= 1, 2, , m, where m is the number of types of equivalent processes, while we designate the set of k type equivalent processes as Fk ={Ps~k}, where s {1, 2, , n}, It is sometimes expedient to have only a single program f or process representation for equivalent processes of a set Fk in a simulation model, where we shall call th is representation the prototype of the k-type processes. We shall call the k-type processes themselves the copies of the prototype in this case. With such an ap- proach to the organization of simulation models, it is sufficient to have m pruto- types in a model, something which tn many cases makes it possible to substantially curtail the volume of a simulation model and a program. Tables of prototypes and copies are used to ref er to a specific process. The num- bers of the prototypes or the types of processes are the values of the table of _ prototypes. The numbers of the copies of the prototypes or the internal numbers of the processes which form the set Fk are the values of the elements of the table of copies. For example, if the process Pi with a number v3 belongs to Fk, then the 98 FOR OFFTCIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY vi-th element of the table of prototypes is equal to k, while the vi-th elemelt of the table of copies is equal to j {1, 2, , e} k The sequence of actions performed by the modeling complex is governed by the organi- zation and functional content of the master program and is conditionally broken down into six steps. In each step, the master program executes the corresponding actions, the major ones of which are: the tranafer of control to the active pracess, the processing of the auxiliary markers, the control of the realization of the coupling of the processes, checking that the conditions for process activatio~ are - met, calculating the time markers, checking for the presence of active processes, searching for the process with the minimal time marker T~in, searching for processes for which the time markers are T= Tmin and correcting the model time. Statistical Data Retrieval and Processing During Modeling. For the sake of brevity, we shall call the set of procedures for sratistical data retrieval and processing during simulation modeling of discrete systems, together with the information base, the system S. This system provides for the retrieval and - processing of data def ined by the user prior to starting the simulation model (prior to the performance of an experiment with the model on a computer). In this case, the user has the capability of controlling the data retrieval and processing process by means of specifying the appr~priate control data: for example, specifying the point in model time, beginning at which the data retrieval and processing will _ be carried out. This capability has been realized for the purpose of providing for data retrieval only after the completion of the transient operational mode of the model, i.e., when the model enters the steady-state mode. Based on the values of the conf idence level and conf idence interval specified by the user, the system S can terminate (at the request of the user) the retrieval of certain data, if an ~ estimate with the requisite precision is obtained. In the case of a limited volume of immediate access memory of t~e computer, in which the experiment is performed, - the user can organize the retr.ieval of the requisite data in a sequential mode. In this case, the retrieval and processing of certain data will be accomplished after the completion, processing and outgut of other data. The system S, in accordance with the information specified by the user, can calculate estimates of the mean values and dispersions of the random lengths of the indicated sections and ttie random lengths of the model time intervals, the bounda~ies of which are determined by the user in the specif ied initial data. The indicated system can also generate histograms of the random interval length and random section length distributions, as well as provide for the capability of routing a requirement with an arbitrary indicated number. In the routing mode, during the modeling process the numbers of the sec~:ions and points in model time of the setting of the indi- cated requirement in a section and its selection from the section are printed out. This makes it possible to trace the motion of a requirement in the model during the modeling process, something which substantially f acilitates the debugging of the models. - The camposition structure and some of the algorithms of the system S have been treated in the literature [6, 7]. 3. The organizat ;onaZ princip Zes of simu Zation mode Zs for CUCC's. Th e M o d e 1 of a CUC C. The computer resources of a CUCC are the computer complexes (VK), 99 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 I FOR OFFICIAL USE ONLY _ tied together by a data transmission network (SPD), formed by data transmission centers and channels. Computer complexes are broken down into mainframe (BVK) an~ peripheral (PVK) complexes. The network centers are designed around com- munication processors, and each computer complex is directly coupled to one of the centers. We sha11 assume that a center, the compuCer complex directly ' coupled to it and the computer complex terminals tied to this form a link. We call all of the elements of a certain link and the users employing the terminals , of this link local with respect to each other, while the elements and users of another link are termed remote,with respect to the given link. A user can interact with the networ'.: in two ways: one-time and an interactive mode, according to the operational mode of the specific computer complex )batch processing or time sharing). In this casn, a user feeds out a message from a terminal, where the addressee is indicatecl in the message. The addressees can be either local or remote computer complexes and users. The message sources can be only the users. Three kinds of inessages come in to each computer complex: internal, external and remote. Internal and external messages come fram local users and are distinguished by the fact that the former are completely processed by the given computer complexes, while the latter are transmitted for processing by remote computer complexes. We shall call external messages addressed to RVK's [mainframe computer camplexes]~the base messages, while those addressed to users are callled communications messages. Remote messages come in from ren:ote users. A computer complex upon receiving a message addressed to it or to = its local user, generates a notification which is roured to the address of the source of the given message. The results of the message processing in the network by one of the computer complexes (local or remote) are routed to the message source in the form of a notice. When switching the packets in a data transmission network, all of the messages and notices incoming to it are formatted as packets of a fixed size. Only the immedi- ate access memory of the co~unications processors is used as the buffer memory at the junction centers. The model of the environment of the CUCC. The environment of a CUCC is created by its users, for whom the category, type and form of the user are defined as the characteristics, where these determine the basic user characteristic: the ranking of the user. Two categories of users are introduced in the models corresponding to a definite external priority level of fihe administrative network, a priority which is assigned to messages of the given users for transmission and processing in the CUCC. There can be two types of CUCC users: one-time and interactive users, corresponding to their modes of interaction with the network, as well as two kinds, determined by - the kind of dominant network resource used during the processing of the user mes- sages. Network resources are broken down into camputational and informational (communications resources are also included in the latter). We then associate . users with the first and second kinds, during the processing of the messages of which, primarily computational and informational network resources resepctively _ are employed, It follows fram the classif ication given here that network users can belong to one of eight ranks. 100 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY Requirements and their attributes. Four classes of requirements are repre- sented in the CUCC models: messages, notices, packats and notifications. They all have a main attribute, the marker, and a supplemental attribute, the complement. The requirements in the models are represented by their numbers, which are generated and erased by the model processes. The number is a unique characteristic, which identifies the requirement in the model. The class of a requirement can change in the process of processing and moving it in the model, but the number of a require- ment always remains constant. The movement of requirements in the model is de~picted ~ by the movement of their numbers. The attributes contain the basic characteristics of a requirement which are repre- sented by indicators. The marker has a composite structure and consists of five indicators, which define the code of the requirement cTass, the number of the prior- ity level (during requirem ent processing and transmission), the number of the user rank to which the given requirement "belongs", the number of the message based on which the requirement was "generated" and the address of the requirement source. The complement consists of five indicators, which define the address of the require- ment addressee, the code for the addressee criterio:~ (this indicator defines and is the addressee of the requirement of the computer complex or user), the number of packets (in the message or notice), the packet number and the packet criterion code (this indicator determines the association of a packet with a message or notice). The markers and complements of requirements are contained in tables of markers and complements. The marker and complement codes are the values of the'elements of the - corresponding tables; the numbers of the elements are equal to the numbers of the - _ requirements. The marker and complement codes are used by and can be corrected by the model processes during requirement processing and transmission in the network ~odel. The structure of the models. The simulation models of CUCC's include four types of program processes which depict de�inite CUCC components: ns depicts the set of users (the environment)of a computer complex, including the users of various categories, types and kinds (the process generates a message flow which depicts the overall message incoming to the computer complex from the users of the given set); ~r~ depicts the computer complex of a CUCC (the process simulates the execution of the main functions of the computer complex of analyzing, proce~sing and transmit- ting the messages, notices and notifications incoming to the computer complex); ~rn depicts a junction center (communications processor) of the data transmission network (the process simulates the execution of the main functions of a data trans- mission network junction center of analyzing, processing, preparing for transmission and transmitting the messages, notices, packets and notifications incoming to the junction); nZ depicts a simplex data transmission channel of the data transmission network (the process simulates the transmission of inessages, notices, notif ications and packets via the channel). The trajectories of movement of the various classes of requirements between the model processes during their functioning depict the trajectories of requirement movement among the components of the CUCC which are depicted by the corresponding processes. The queues which are realized in the models based on the principle of processing the requirements located in them are broken down into dynamic and static types 101 FOR OF~ICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY (queue listings). Dynamic queues are the input queues to~~the~processes, while static queues are the internal queues of the processes. The requirements from a dynam3c queue are procesyed as processes without interruptions, i.e., until all of the re- quirement found within the queue are processed. The processing of requirements from a static queue can be interrupted, specifically: upon completing the processing of the next regular requirement, a process can terminate the processing of the require- ments from a given queue and changQOVer to the performance of other functions, in particular, begin the processing of requirements from other queues. Setting and selecting the requirementsfrom the queues is accomplished in accordance with the "First come - first served" principle, with relative priority, i.e., without inter- rupting the procPSSing of a lower priority requirement upon the arrival of a higher priority requirement. The queues in the model are realized in the form of coupled listings. 4. SimuZa tion modeZing of a CUCC. Some results of a study using simulation modeling of one of the organizational variants of a CUCC, which we shall call the VTsKP-G [CUCC-G = Collective-Use Computer Center - G] are treated in this section [8, 9]. The CUCC-G includes 13 computer complex- es: 10 peripheral and 3 mainframe com- z~ Z,~ Z17 Z~~ z~o plexes. All of the peripheral c~mputer complexes are based on the M-7000 com- puter, while the base mainframe computer Zr Z9 complexes, the BVKYe, BVKB and BVKE, are based on the Unified Systezn of Electronic Computers, the BESM-6 comput~r and the - z3 z+ Zs Z6 zi zs "E1~~~::~" muTtiprocessor computer. All of the computer complexes are tied to- gether by the data transmission network. Figure 1. The topology of the data The functions of the data transmission trunsm ission network of the CUCC-G network junctions in the CtJCC-G are partially realized b,. the M-7000 com- puters incorporated in the computer complex, and partially by r,.~ ro~~mputers. The topology of the data transmission network is shown in Figure 1; -~he~e Zi, i= 1, 2, 13, dcsignates the links of the CUCC-G, while the lines cc~~espond to duplex data transmission channels, each of which is designed in the fc~~ uf two simplex daCa transtaission channels (KPD) , one of which is the input cha �.ie1 �relative to a specif ic junction, while the other is the output. The carrying capacities of all data transmission channels are : b Kbit/sec. The iinks Zi, 1, 2, 10, correspond to the peripheral compute c,~mplexes, while Z11~ 'l.~2 and Z13 are the BVKYe, BVKB and BVKE respectively (therE ~rF no users having direct access to the BVK [base mainframe computer complex, ~-~:~sr:, .,ers in the links corresponding to the BVK's). The major algorithms and sa.:. >f. che parameters of the CUCC-G and its environment were treated in [10]. The CUCC-G was studied through experiments with simulr.'', k~:r mc~..`als: the IM-2 and IM-3. The IM-2 model depicts the CUCC-G as a whol, ,~~ni~.z p.:~� IM-3 model depicts the CUCC-G data transmission network. Various modiflca~ic:r~', rf the processes ns, nc= nn and nZ are used in these models. The proces,~.:. is1 ~ea~,~~. models are combined into blocks. The block Bi, i= 1, 2, 13, dep:cCs -:hc ;ink Zi of the CUCC-G 102 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFF[CIAL USE ONL'~ m �M o ~ . ~t,i ,-I a ' ~ N ,,G 'a `a ' ~`o u 'v u~ N ~ 1 ~ o o ro .~c ~ o ~ ~ � ~ ~ ~ . .'~I,i ~ ~ ~ ~ ~ ^ N ~ ' ~ l ~ ~.C ~ ~C,i y~i ~ ~ ~ r,~~~ ~c y r,r o ~ ~ o . ~ ~;t . Figure 2. The configuration of the Figure 3. The configuration of the block Bi of the IM-2 model block Bi of the IM-3 model. with the output data transmission channels. The blocks Bi in the IM-2 and YM-3 models differ in terms of the composition and content of the processes included in - them. One each of the type ~rS, ~r~ and ~rn processes is incorporated in th~ Bi of the IM-2, as well as ri type ~rZ processes, where ri > 1. Incorporated in the Bi of the IM-3 are the same types of processes, with the exception of ~r~, the basic characteristics of which are depicted in modifications of the ~rs and nn processes. The structural configurations of the Bi blocks in the IM-2 and IM-3 models are shown in F igur es 2 and 3. One of the systems analysis problems solved by means of simulation modeling was the study of the impact of various methods of organizing data transmission via the data _ t~ansmission link on the quality of CUCC-G functioning: message switching and packet switching. The results of the relevant experiments with the models are treated be- low. An algorithm was realized ir the models for routing the messages and packets, which provided for their transmission between the links of the CUCC-G (between the centers) via minimal routes (with the least number of interm.~diate ~unctions). If several minimal routes exist for the transmission of inessages and packets from Zi to Z~, j~ i, i,j = l, 2, 13, then they are transmitted with equal probability via one of them. The addressing of external messages to various CUCC-G computer com- plexes was specified in the modeling by the distribution of the probabilities d= _(d0, dl, d2, d3), where dp, dl, d2 and d3 are probabilities of the addressing of external messages to the following respectively: the peripherll computer complexes ~ (users), th e BVKYe, BVKB and BVKE. A distribution of d=(0.03, 0.93, 0.02 and U.02) was specif ied in the experiments with the models and the existence of two categories of inessages was depicted in the CUCC-G (two categories of users) corresponding to the level of their external priority (20 percent of the messages were of the fi.rst = category, and 80 percent were of the second). It was presupposed in the switching of the packets (the packet size was fixed) that in the case of incomplete f illing of a packet with data, it ~oas filled completely with "zeros", the transmission of which produces an additional load on the data transmission channel. This way of depicting the transmission of incomplete packets was used to partially take into account the additional loading of the data trans- mission channel produced by the repeated transmission of packets in which errors were detected which were caused by interference in the channels, since the actual 103 _ FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFIC[AL USE ONLY mechanism for detecting errors in packets and providing for repeated transmission of packets received with errors which is present in the CUCC-G was not reflected in the models for the sake of simplicity. Considering the fact that dl � dZ,d3, the volumes of the base messages in the BVKB and BVKE in the experiment with the IM-2, the sizes of the base messages ad- dressed to these BVK's and the sizes of the notices generated by these complexes were all specified as equal to the f irst moments of the corresponding distribution _ functions. For the purpose of reducing the time required to lock the model into ' the steady-state mode when generating the initial state of the model (during each experiment), the initial setting~of the messages was accamplish ed in the queue of messages awaiting processing in the BVK, where the number of inessages was equal to the average number of inessages in these queues, determined by means of analytical modeling [11]. The fact that the average length of time the messages were present in the base computer complex of the CUCC-G amounts to about 52 seconds was taken into account - in the experiments with the I~I-3 when specifying the initial data for the modeling. This value was obtained by means of analytical modeling of the CUCC-G and was con- _ firmed in simulation modeling using the IM-2. The average response times of the CUCC-G for external messages and the load factors of the devices were taken as the basic parameters of the CUCC-G, which characterize its operational quality. In particular, the average response times for base mes- sages, especially messages addressed to BVKYe's, were of the greatest interest, _ since the utilization of the problem oriented program systems of an information and reference nature which were realized in the BVKYe and which are under development at the present time, was depicted in the modeling. A study of the response times for base messages was made by means of the IM-2. Also taking into account the f act that the switching method basically influen.ces the time characteristics of data transmission through a data transmission channel, it was important to determine the extent of this influence. For this reason, the average transmission through a data transmission system, which is the sum of the average duratic+n of time that a mes- sage and the notice corresponding to it are present in the data transmission system was studied as a f unction of the switching method and packet size (in the case of - packet switching) in the experiments w~ith the IM-3. We shall introduce symbols for the estimates of the average response times of - the CUCC-G and the load coeff icients of the data transmission channel, in which 4 S= Ye~ B and E is the indicator of the BVR, k= 1, 2 is ~b~ indicator of the categ:~ry (t:he external priority level) of the messages: p~ is the estimate of the ~b~i,~^~ response time of the CUCC-G for base messages, addressed to the BVKS; pk~s is the estimate of the average reaponse time of the CUCC-G for k-th category base messages, addressed to the BVKS; p is the estimate of the average transmission time through the data transmission system; pk is the estimate of the average transmission time through the data transmission system for k-th = category messages and notices; ~(i~j) is the estimate of the average load coefficient of a data transmission channel, which is the output channel for the i-th junction center and the input one for the j-th junction. 104 FOR OFFICIAL USE ONLY APPROVED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 APPR~VED FOR RELEASE: 2007/02/09: CIA-RDP82-00850R000400020012-2 FOR OFFICIAL USE ONLY The results of twq2~eri ~~of ~periments are considered below, th~ first of. which (the experiment J~ , J~ , J~ 3,was p~rformed using the IM-2 model (see Table 1), while the second (experiment ~~3 was performed with the IM-3 model (see Table 2). In eperiments ~i~3 and , a message switching method was TABLE 1 Ta6nNUa 1 BVKYe 6BKE gp~ ssxs I SBK3 $~Jj~ I 3i is~ I ~ cs> I 3~ t2~ I 3i ctl . 3~ . 3, 3i . 3: . 3, . p~cb~ (c) I 35 I 41 I~l3 I 900 I 400 p~.