A FORMAL METHOD FOR SCHEDULING ANALYSIS OF A PARTITIONED MULTIPROCESSOR SYSTEM - Dynamic Priority Time Petri Nets

Abstract

In order to examine whether the timing constraints of a Real-Time application are met, we propose an extension of Time Petri Nets model that takes into account the scheduling of a set of tasks distributed over a multiprocessor architecture. This paper is concerned with dynamic Priority-driven scheduling, whose policy is known to be supported by a new formalism called dynamic Priority Time Petri Nets (dPTPN). Its ultimate objective is to show how to deal with the Least Laxity First Scheduling policy with a set of periodic independent tasks. Besides, the use of dynamic priorities gives a determinism aspect to the model in which a crossing of concurrent transitions exists. Therefore, the execution of the model is accelerated and the number of accessible states is decreased.