A novel PSO algorithm for dynamic wireless sensor network multiobjective optimization problem

Abstract

All real-life multiobjective optimization problems (MOPs) are considered dynamic. It is occurring due to fluctuations in environmental or the global market instabilities that lead to the quick fluctuations of prices. Wireless sensor network optimization problem, by nature, is considered one of the dynamic MOP (DMOP) and needs to be solved by a special method to save time and effort. Thus, this paper proposed a novel particle swarm optimization (PSO) for dynamic wireless sensor network MOP to accelerate data transfer in networks and reduce energy losses. Generally, in DMOPs, the optimization period is broken into several equal subperiods. In each subperiod, there is a dynamic parameter that changes. In the proposed approach, PSO is used to handle DMOPs without any changing of its structure and has fast convergence properties. However, a new mechanism to choose the personal and global preferred particles is introduced, which based on the distance between the nondominated solutions obtained so far and the particles positions by using Euclidean metric. In addition, two types of archives are used to maintain the nondominated solutions. The first one is used to store the nondominated solutions obtained by each particle, whereas the second type is used to store the nondominated solutions achieved by all particles. The novel methodology performance is proved by applying it on three benchmark problems that were chosen from the literature and one design problem from the engineering domain. The simulation results mentioned that the proposed algorithm is active and efficacious in solving DMOPs.