Design of a learning fuzzy production system to solve an NP-hard real-time assignment problem

Abstract

This paper presents the design of a Learning FuzzyProduction System to solve an NP-hard problem consisting of a set of real-time preemptible tasks to be assigned to a set of heterogeneous processors, with placement, resource, communications and time constraints. Time to obtain the first solution and the number of solutions found for d@erent known problems are given.Real-time scheduling is one of the areas in which efforts must be concentrated in order to develop a science of large scale real-time systems [SI. The problem of assigning a set of preemptible real-time tasks to a multiprocessor system is NP-hard. It has been attacked by heuristic methods [4, 7 , 11, 121 among others, and by simulated annealing [l, 101. In this paper, a Learning Fuzzy Production System (LFPS) to solve the problem is presented.2.

Power saving and fault-tolerance in real-time critical embedded systems

“…The assignment problem is NP-hard and in the past has been addressed by many methods belonging to the partitioned class, e.g. Heuristics [36][37][38], Simulated Annealing [39], Genetic Algorithms [40] and Fuzzy Algebras [41].…”