A Multi-Agent Based Hybrid Optimization Method for Signal Source Search and Localization

Abstract

Searching and locating the position of the signal source is of great significance in wireless sensor network, mobile communication, public safety, and so on. A traditional signal source search and localization method requires either the signal source to be in the line-of-sight or a large number of operations to compute, which is difficult to satisfy in a complex environment for an unknown signal source with limited computational resources. In this paper, we propose a hybrid gradient-free optimization method combining the advantages of the Nelder-Mead Simplex algorithm and the Particle Swarm Optimization to study the search and localization of a signal source by using received signal strengths with a multi-agent system. Integrating a direct search method with a bio-inspired evolutionary method enables a feasible optimal solution to be found with a rapid convergence rate. To validate our proposal, numerical experiments are conducted to investigate the localization performance; three cases are studied, including two standard objective test functions and a complex 2.4GHz mobile signal strength distribution. The findings demonstrate the approving achievements of the proposed method in terms of global optimization, accuracy, and rate of convergence.