Particle Swarm Optimization Based Particle Filter Techniques for Target Tracking in Multistatic UWB Radar Sensor Network

摘要

Sensor networks are mainly used for applications such as emergence detection, target of interest monitoring, non-cooperative human object detection such as an intruder (target) for border surveillance, intrusion unauthorized movement around critical facilities. In this thesis, two new algorithms are proposed for target (human intruder) tracking in an Ultra WideBand (UWB) multistatic Radar Sensor Network (RSN) consisting of one transmitter and multiple receivers. These algorithms are based on Particle Filter (PF) with embedded variants of Particle Swarm Optimization (PSO) techniques to provide a solution of tracking problem in dynamic and noisy environments. First algorithm; Adaptive inertia Weight Particle Swarm Optimization Particle Filter (AWPSOPF), is a PF embedded with fitness based adaptive inertia weight PSO that improves the convergence of PSO, tackle PSO bias issue, solves PF sample impoverishment problem and improve tracking accuracy. The second algorithm; Distributed Particle Swarm Optimization Particle Filter (DPSOPF), is an enhanced version of AWPSOPF where distributed PSO is embedded in PF and PSO particles are divided into further small groups based on minimum distance which provides a robust solution for target tracking problem.