IFSR Newsletter 1986 Vol. 6 No. 3 Winter
A short report on a research activity at the Institute of Systems Science of the Johannes Kepler University Linz, Austria
Systems Theory is a field of applied mathematics which was developed to help engineers deal with models at different levels of abstraction. It is only about 40 years old and is still undergoing rapid transformations. It might be considered as having arisen out of control theory and signal theory. A large number of articles concerning systems theory appear in various scientific journals. Many monographs, textbooks, international conferences and university courses also deal with this field of science. It might therefore seem that systems theory has made amazing progress and that its devotees should be very satisfied. Unfortunately, however, the degree of acceptance of systems-theoretical methods in practical engineering is low; only a few of the most important of these techniques are now generally used as tools in the engineering workshop. The reasons for this are manifold. One of the main ones appears to be the lack of powerful and easy-to-handle computer assistance for the application of the systems theoretical framework and methods. Many of the most important results are only described in textbooks; for their study it is necessary to resort to “paper and pencil I methods”. The software into which algorithms have been designed is usually not suited for use with the types of computers currently being employed by engineers. Therefore a new implementation would be necessary, but it is usually considered too costly; the consequence is the exclusion of systems theory from practical application. On the other hand we must also take into consideration the fact that the usage of mathematics and of systems theory requires that the engineer have some understanding of abstract models. Theoretical insight is also necessary if he is to be able to select the proper methods. Although he has certainly acquired such knowledge during his engineering studies, he is likely to have forgotten it very rapidly after graduation. Engineers remember primarily those mathematical concepts for which there are practical applications.
There have been a number of projects at different universities to realize a computer-assistance of systems theory for various purposes. As an important basic example we shall mention the “general systems problem solver” (GSPS) implemented by Professor George Klir and his research group at the State University of New York at Binghamton, USA. GSPS is a rather complex software package for general use. It realizes some of the most important parts of the general systems methodology as described in numerous articles and books on this subject. Another exam pie of the more general part of systems theory is given by the work of Professor Bernard Zeigler of the University of Arizona, Tucson, USA, which realizes a multilevel systems design and simulation methodology. It can be used for the engineering task of any systems design which requires a multifacetted approach.
There are many other examples for the implementation of systems theory methods in software. Most of them are constructed for special tasks, such as controller-design, electrical network synthesis or digital filter design It is the goal of our group to help promote a general approach for the construction of systems theoretical tools for design and simulation. We propose that modern computer languages such as LISP or PROLOG be used for functional programming and we especially emphasize RAPID-PROTOTYPING. In this way we also encourage the user to implement his personal routines for problem solving as parts of the program-package.
In our programming task we want to make use of not only the “artificial intelligence” languages LISP and PROLOG but also of the mechanisms which are provided by expert systems. Therefore we are interested in being able to employ an expert system shell, such as for example LOOPS or KEE, to offer the user enough comfort for data and method manipulation. In ‘”Our case we use the L1SP dialect INTERLlSP-D which provides additional support for interactive work and display graphics together with LOOPS. All our CAST-systems are implemented on SIEMENS/XEROX Dandelion work stations which are interconnected by means of the ETHERNET.
We are still in an early phase of systems development. We are using a basic framework called STIPS (= Systems Theory Instrumented Problem Solving) which is similar to the framework used in GSPS by Professor Kill’ and to the framework of Professor Zeigler’s “Multifacetted MotJellinil and Simulation Methodology”. By specializing the STIPS framework to subclasses of systems specifications and related design and analysis methods we obtain method-banks which can serve as tools in engineering workstations. One project in which we are involved is designated as CAST:FSM.
That stands for “Computer Aided Systems Theory with emphasis on t.he case of Finite State Machines”. It is intended to provide the user (usually thought to be a Logic Designer involved in computer systems design) with the “world” of automata theory which is relevant for hardware problems. Our goal is to devise a system of programs (implemented in INTERLISP/LOOPS) to enhance current computerized tools of logic design workstations. Another project in which we are currently involved concerns”Design for Testability” problems in VLSI-circuitry design. In addition to the special parts of CAST:FSM which can be used for that type of problem (e. g. for structuring finite state machines to become more easily testable) we try to develop new methods for functional testing (= testing from a user-information point of view) and for fault-simulation (rule-based simulation procedures based on the language PURLOG, a PROLOG extension). Furthermore, within this project we are developing the tool L1SAS for systolic array testing (L1SAS = Linz Systolic Array Simulator). Finally, we want to adapt the CAST philosophy so that it will apply to other problem areas such as cryptography (design of hardware for data encryption) and digital image processing: both of these fields could very well be aided by systems theory.
Developing CAST tools for the synthesis and analysis of systems is one of the most important goals of the systems scientist. Pioneering studies in this field have already been published by Prof. George Klir (SUNY Binghamton) and his colleagues and by Prof.
IFSR Newsletter 1986 Vol. 6 No. 3 Winter