Showing papers on "Process architecture published in 1974"

Analysis of asynchronous concurrent systems by timed petri nets

Abstract: This thesis is concerned with the modeling and performance analysis of systems which consist of concurrently acting components, an example of which is an asynchronous pipelined processor. The work is divided into two parts. In the first part, a suitable model is developed for describing the structure of asynchronous concurrent systems. In conventional automata theory, the finite-state machine model is used to describe the behavior of systems; the problem with this is that a large number of states results when practical systems are modeled. In this thesis, each system component is modeled as a finite-state machine, and a system is viewed as an ensemble of interconnected finite-state machines. This has the advantage that the size of a system model grows linearly rather than exponentially with the number of system components. A subclass of Petri nets known as SMD (State Machine Decomposable) Petri nets is identified in order to formalize the notions of finite-state machines and their inter-connection. For convenience, systems of interest are divided into two broad categories: (a) deterministic, or decision free, (b) non-deterministic, or systems with decisions. SMD Petri nets are used to model both classes of systems; in addition, a subclass of Petri nets known as LSP Petri nets is used to model those deterministic systems that cannot be modeled by SMD Petri nets. The second part of the thesis is concerned with finding the computation rate of activities in real-world asynchronous concurrent systems. Practical systems are constructed from devices which have a finite speed of operation. Since Petri nets do not have time parameters as part of their definition, they can model the structure of systems but cannot be used to study their computation rate. The definition of Petri nets is augmented to model the speed of operation of a device in a system by assuming that the corresponding activity in the Petri net has a finite, non-zero time duration. The resulting nets are termed timed Petri nets, and methods are given for finding the computation rate of activities in times SMD and LSP Petri nets. The results are applied to the analysis of several asynchronous systems drawn from areas within and outside the domain of computer systems.