Classified Particle Swarm Optimization Based Algorithm for Cooperative Localization

Abstract

High-precision wireless localization has attractive application prospects. Cooperative localization is an effective tool to improve localization accuracy. However, compared with non-cooperative localization, in cooperative localization networks, large-scale neighboring links and nonlinear measurement functions cause the associated objective function to be non-convex. It is difficult to obtain global optimum using classical particle swarm optimization (PSO) algorithm or analytical methods. In order to solve this problem, a classified particle swarm optimization (CPSO) algorithm is proposed in this paper. For classical PSO, all search particles have the same inertial weight and learning factor. Unlike classical PSO, the proposed CPSO algorithm classified different search particles based on particle cost value and set different inertial weights and learning factors for search particles. Meanwhile, considering the unavoidable reference node location error, localization result could be achieved by calculating the weighted average of close-range particle locations. Simulation results prove that the CPSO algorithm improves positioning accuracy by 25.3% compared with classical PSO algorithm.