Bat intelligence search with application to multi-objective multiprocessor scheduling optimization

Abstract

In this paper, we present the bat intelligence search for the first time. Bat intelligence is a novel and unique heuristic that models two major prey hunting behaviors of bats: (a) utilization of echolocation to observe the environment and (b) employment of constant absolute target direction approach to pursue preys. In order to illustrate the performance of bat intelligence, we implement this heuristic to solve two types of multiprocessor scheduling problems (MSP): single objective MSP and multi-objective MSP. In single objective MSP, we independently solve for minimization of makespan and minimization of tardiness. In multiple objective MSP, these two objectives are optimized simultaneously. In the single objective MSP, on average, the bat intelligence outperformed the list algorithm and the genetic algorithm by 11.12% when solving for minimization of makespan and by 23.97% when solving for the minimization of tardiness. In comparison to the genetic algorithm, the bat intelligence produces better results for the same computational effort. In multiple objective MSP, bat intelligence is combined with normalized weighted additive utility function to generate a set of efficient solutions by varying the weights of importance. The results demonstrate that the bat intelligence finds a set of Pareto optimal solutions on bi-objective optimization of MSP.