Bearings-Only Tracking of Manoeuvring Targets Using Multiple Model Variable Rate Particle Filter with Differential Evolution

Abstract

In standard target tracking, algorithms assume synchronous and identical sampling rate for measurement and state processes. Contrary to these methods particle filter is proposed with variable rate. These filters use a restricted number of states, and a Gamma distribution is applied at state arrival time so that the maneuvering targets could be tracked. Although this structure is capable of tracking a wide range of targets motion features using linear, curvilinear motion dynamics, it suffers from a basic weak point. It cannot estimate the position of targets in high maneuvering regions. Thus, multiple model variable rate particle filter (MM-VRPF) is utilized to overcome this shortage using various dynamic models. A weak point of a particle filter is a phenomenon called degeneracy that even exists in MM-VRPF structure. In this study, differential evolution method, is exploited to improve the mentioned method and a novel structure called multiple model variable rate particle filter with differential evolution (MM-VRPF with DE) is introduced. The simulation results of a bearing only tracking achieved from a maneuvering target, revealed that the proposed structure has better performance while it maintains advantages of variable rate structure.