Static Task Scheduling Using Genetic Algorithm and Reinforcement Learning
Mohammad Moghimi Najafabadi, Mustafa Zali, Shamim Taheri, Fattaneh Taghiyareh · 2007
Task scheduling in a multiprocessor system is defined as assigning a set of tasks to a set of processors. The goal is to minimize the execution time while meeting a set of constraints. A wide variety set of deterministic and heuristic methods are proposed to solve the problem. The main problem is that the proposed methods cannot deal with big search spaces and cannot guarantee to find the optimal solution. In this research a novel approach based on reinforcement learning and genetic algorithm is proposed. Being divided using genetic algorithm, the smaller problems can be solved with reinforcement learner scheduler. The result of the method is a set of task processor pairs. Simulation results in standard problem set show that the method outperforms some studied GA based scheduling methods