Designing cellular automata-based scheduling algorithms
Franciszek Seredyński, Cezary Z. Janikow · 1999
In this paper, we present a systematic approach to designing cellular automata- based algorithms for multiprocessor scheduling. We consider a simple case of two processors. However, we do not restrict parallel programs. We show how to design local neighborhoods and corresponding cellular automata (CA) for a given program graph. We also show how to discover, by genetic algorithm (GA), rules of CA - suitable for solving a given scheduling problem. We evaluate the discovered rules in terms of applicability to cope with different instances of the scheduling problem.