Directed Search-based PSO Algorithm and Its Application to Scheduling Independent Task in Multiprocessor Environment

Abstract

Particle swarm optimization (PSO) algorithm has proved to be a promising meta-heuristic algorithm to solve broad class of optimization problems which requires global search. Many variants of basic PSO have been proposed. To enhance the exploration capacity of basic PSO algorithm, a new technique called as directed phase is introduced in PSO. The proposed new phase is based on directed search optimization (DSO) which has capability of exploration and diversification which can accelerate the particles in the late iterations of PSO algorithm. Further, proposed algorithm along with PSO and DSO is implemented to solve task scheduling problem on homogeneous multiprocessor system and the results obtained are compared. Experimental results demonstrate that proposed work performs better and has the ability to be an adequate alternative to solve the optimization problem.