Real-Time Systems

Since 2004, most of chip vendors have begun to shift their major focus from single-core to multi-core architecture (W. Wolf. Signal Processing Magazine, IEEE, 26(6):50– 54, 2009). One major reason of this shift is that it reaches a physical limit by scaling transistor size and increasing the clock frequency to improve the computing perform‐ ance on a single-core architecture (Agarwal et al. Proceedings of the 27th Internation‐ al Symposium on, pages 248–259, June 2000), that is, the overall chip cannot be reached within a single clock cycle. Multi-core architecture, however, brings innovative and promising opportunities to further improve the computing performance. By provid‐ ing multiple processing cores on a single chip, multi-core systems can dramatically increase the computing performance and mitigate the power and thermal issues with the same performance achievement as single-core systems. As multi-core architecture has been more and more dominant in the industrial market, there is an urgent demand for effective and efficient techniques for the design of multi-core systems. In this chapter, we first analyze the thermal behavior on multi-core real-time systems by taking the heat transfer among different cores into consideration. Then we analyze the energy consumption for a given speed scheduling on multi-core systems.

0~B 7 G_~IT--An Interactive Graphics Display System. J. Ballance, J. Baughman, and L. Hubble, Oregon State U.--The availability of efficient time-sharing services and low-cost graphics terminals has opened a new and exciting approach to problem investigation not only for researchers, but also for students in undergraduate education. A highly interactive graphics system has been developed at Oregon State University which provides graphic or numeric output at the user's option. The system has capabilities for handling: A) arrays of data, B) functions specified by a formula, C) systems of differential equations, D) systems of difference equations and E) special numerical computations such as integration, differentiations, etc. In addition, the user may at any time define his own command by supplying a FORTRAN subroutine to do the desired operation. For example, the user-defined commands may be for filtering, spectral analysis, creating special displays such as perspective views of surfaces, contour plots, or any other operation specifically tailored to the requirements of the user.

U2~The
Cell Display Processor for Line Drawin$s. BERNARD W. JORDAN, JR., Northwestern --A special purpose processor capable of generating arbirtrary line drawings in "real time" is described.
The feasibility of such a display is shown by deriving the minimum number of patterns required to synthesize an arbirtary line. The synthesis process does not compromise picture quality, since the resulting dot patterns are identical with those of a conventional raster display. Furthermore, the time constraints of a raster display are shown to be satisfied for a typical design for very complex line drawings.
~ usin$ Interactive Graphics to Obtain Approximate Solutions to Partial Differential Equations. PAUL LAFATA, University of Wisconsin --An interactive graphical system which can be used to approximate functions of two variables or to obtain approximate solutions to partial differential equations is described. The function or differential equation is entered symbolically, so the system may Be used by someone with no programming experience. The basic mathematical technique is to assume the approximate solution is a linear combination of user selected functions, and to use linear programming to determine the coefficients of these functions, l Contour maps and one-dimensional crosssectional plots of the approximate solution and error information can be graphically displayed as requested by the user. Based on these results various parameters may be modified on line so as to obtain a better approximation. Variable coefficient elliptic, hyperbolic, and parabolic partial differential equations have been approximated to Michie's memo function proposal I allow this technique to be effectively extended to real time evaluation of functions of (single) real arguments on a small digital computer.
A description of the new algorithm is followed by a report of the results of a series of experiments which demonstrate notable increases in computation speed for repetitive evaluation of two functions on a PDP-9 computer. Argument data for these experiments were randomly generated i000 element sets with normal, exponential, and uniform distributions.
A further experiment investigated the dynamic response of the memo dictionary for an argument population with time varying parameters. Some comparisons are drawn between replacement of dictionary entries in a memo function and page replacement algorithms in virtual memory systems. Memo functions provide a means of retaining historical information concerning the performance of real time processes.
Such data when treated statistically may be used to assess the validity of input variables to which decision logic then can be applied before returning output control values. The adaptive nature of memo functions is applied to batch processes by controlling changes in setpoints as the reaction proceeds and also by tracking the variations of controlled parameters in unstable reactions.
Implementations are discussed for both dedicated and multiprogrammed minicomputer systems.

W~S
A Possible Distributed Computing Network. ALAN MINK, U. of Md./NAVSEC --In an effort to increase resource sharing, provide ease of modular expansion, and facilitate computing availability, a distributed computing network is proposed. This network is a collection of computers and resources that are physically independent, yet operate under a unified line of control that does not reside in any one physical unit, but is distributed throughout the system. The network configuration appears as a set of concentric rings whose center is a file system. Moving outward the other rings are: (I) the file processing, high speed communication ring, (2) a ring of processing nodes (each node is an independent processing system),(3) a slow speed communication ring, connecting (~) a ring of peripherals through which the user interfaces. Each processing node will contain its own operating system, while the network itself will have a distributed operating system which will handle spooling, job and resource allocation, locating and passing files, and interprocess communication.
The composition of any individual processing node can vary from a single minicomputer to large multiprocessing systems. The processing nodes of the network are not necessarily homogeneous, their number and composition will depend upon the intended computing loads. U.7~e~ A Methodology for Communication Program -DAVID L. FULTON, Bowling Green State -This paper describes the techniques used in an implementation of the Binary Synchronous Communication procedure for point-to-point telecommunications.
While developed for this specific protocol, the techniques are general and applicable to any other communications procedure. The essential notion is to rely on the mechanisms of formal language and automata theory to arrive at a formal description of the communication discipline.
Then, given such a description, it is possible to apply standard techniques of automata theory to arrive at an automaton which is the basis of the syntax and semantic analyzer through which communication is controlled. This technique provides a simple, mechanical way to construct a communication program which can, moreover, be easily altered to fit changing needs.

~C~A
HEALTH SCIENCES COMPUTER NETWORK. ARNOLD G. NEL-SON,MICHAEL W.DIFFLEY,UNIV, OF MINNESOTA--The rationale, structure, and implementation of a computer network which employs a queued message switcher implemented on a PDP-8 minicomputer is summarized. Each network terminal may communicate independently with any other terminal. The major computational resource on the network is a Control Data 3300 system which is connected to the PDP-8 via a channel-to-channel inter-computer coupler (ICC). Other network terminals consist of" teletypes, teletype-compatible CRT's, and laboratory minicomputers communicating over telephone lines at speeds of ii0 to 1200 bits per second. The primary use of the network, at present, is to provide remote access to the CDC 3300. This allows interactive use of a medium-large general-purpose computer with considerable on-line file storage capacity. Synchronous phone-line transmission techniques will be implemented shortly on the PDP-8 to facilitate the development of remote job entry stations (RJE's) which will access the 3300 at speeds of 2400-9600 bps. The central communications approach provides for a flexible and relatively inexpensive research tool in the study of network techniques applicable to Health Computer Sciences at the University of Minnesota.

~.~Design
and Certification of a Secure Colmnunications Processor.* PETER TASKER, The MITRE Corporation --An experimental message processor (MP) was designed to operate in a multi-level classification environment using packet-switching network disciplines.
Early attention to security access controls has resulted in a system that minimizes the amount of software whose correctness must be certified.
Most system parameters (line speeds, number of terminals, interactive dialog for terminals, communications line disciplines, massage format) are under the control of application software that may be user written but cannot cause security compromise.
The MP uses two processors tailored through firmware to realize the functions of a multi-level input/output processor and a multi-level control or central processor.
Run-tima controls implemented in firmware are invisible to the applications program unless it attempts a security violation. Attempted violations and system error conditions are logged in a security control process that can observe and, if necessary, terminate offending processes.
Techniques have been developed for certifying as correct those parts of the system critical to preventing compromise. *This work is being sponsored by the United States Air Force under Contract Number 19628-73-C-0001.
~J~C 7 Computer Systems Analysis. EDWARD K. BOWDON, SR., University of Illinois --This research is aimed at developing analytical tools for system modeling and analysis of real time computer networks.
We have begun by formulating a geographically distributed network of computers.
From a postulation of the essential characteristics of our computer network, we have developed a queueing theory model for a multiserver system with a finite length priority queue. Then, under the assumptions of Poisson input and exponentially distributed processing times, we utilize this idealized mathematical model to investigate the steady state stochastic behavior of jobs in the network. We are particularly interested in the efficiency of computer utilization and the average waiting time for jobs of different priority classes. Additionally, we have employed the average waiting time results from the queueing theory model to determine steady state job dispatching rules for a network of centers. From a knowledge of the average arrival rates and processing times for each priority class, we determine what fraction of the jobs in each class should be transmitted between centers in order to balance the average waiting times for each priority class throughout the network.
(~-~S FDM/FM Spectral Analysis. R. E. RUSSEI,I_, and J. S. ROCHEFORT, Northeastern U.--The computation of the power spectral density of a radio wave which is phase or frequency modulated by a sum of sinusoids has been investigated for the special case of subcarriers that are widespread in frequency. The spectrum was evaluated by a computer program which implemented the Middleton expansion, a weighted series of multiple-fold self-convolutions of the phase rnodulating spectrum.
This method is shown to yield good results when compared with the classical infinite-sums-of-Bessel-functions solution in cases where the latter approach requires prohibitively long computation time for convergence.
A numerical convolution approximation and computer-imposed constraints on the Middleton expansion are discussed for frequency division multiplex telemetry systems using the common IRIG subcarrier set.

O~l~ ~
Easily Implemented Compiler for Real-Time Control Minicomputers. RICHARD L. ROTH and EUGENE HUNTING, Univ. of Pittsburgh --A compiler for a subset of FORTRAN was constructedlusing the implementation language BLISS-10 , running on the DEC System/10. Producing assembly code for the PDP-7, the compiler was written in less than three manmonths.
The design is such as to facilitate easy extension, so that specialized FORTRAN subroutine calls can be translated into partially optimized segments of minicomputer real-time control programs.
The modular construction of the compiler is also designed to facilitate alternative coding modules, for other object machines.
Implications for the software engineering problem will be discussed.