Affiliations: Advanced Digital Broadcast Polska Ltd., Zielona
Góra, Trasa Północna 16,65-119 Zielona Góra, Poland | Institute of Computer Computation Engineering,
University of Zielona Góra, Podgórna 50, 65-246 Zielona
Góra, Poland | Department of Telecommunications Technical University
of Koszalin, Racławicka 15-17, 75-620 Koszalin, Poland | Institute of Engineering Cybernetics, Wrocław
University of Technology, Janiszewskiego 11/17, 50-372 Wrocław,
Poland
Abstract: This paper is addressing an issue of distributed systems designing
aimed at automated prototyping of Cyclic Concurrent Processes Systems. In such
systems concurrent processes compete for access to shared system resources. In
order to ensure that a system is deadlock and starvation-free, certain
conditions must be satisfied. In this paper, these conditions guarantee that
for a given pair (an initial state, a set of dispatching rules) the system -
belonging to a specific class - has a steady cyclic state. However, system
designers are interested in values of performance indices, such as a rate of
resources or processes utilization or the period of the system cycle. Nowadays,
the values of performance indices are provided mainly as a result of a
simulation process, which requires much more processor power than in case of an
analytical method. Thus, in this paper the authors focus on providing a
procedure that enables building analytical models of Cyclic Concurrent
Processes Systems belonging to a system class considered in the paper. To reach
this aim the max-plus algebra formalism is employed. Both the conditions
ensuring a cyclic process flow and steps of the procedure are the basis of a
software tool, which can be used by designers to prototype systems of desired
values of the performance indices. Thanks to the computer program the designers
receive a useful tool that helps to validate and allocate distributed control
procedures, even in a complex system, which is a composition of simpler
systems. The procedure together with the software tool is the main outcome of
this paper.
