DYNAMIC TUNING OF PARTICLES NUMBER IN A PARTICLE FILTER USING FUZZY LOGIC SYSTEM TO TRACK HIGH-PERFORMANCE TARGETS

Abstract

To solve the problem of nonlinear non-Gaussian state space model engendered due to tracking high-performance targets, the particle filters were used. Obviously, a large number of particles provide more accurate calculations for the posterior density, which results in a better tracking performance, especially when tracking a high performance target. Unfortunately, a large number of particles results in more computation load. Therefore, we need a dynamic tuning of the particles number to provide a relative small particles number for a non-maneuverable target, and a relative large particles number to track a maneuverable target. In this paper, a new fuzzy logic system for dynamic tuning of particles number in a particle filter is introduced. The fuzzy logic system is used to choose the suitable number of particles based on the maneuverability of the target of interest. It assigns a large number of particles to track high-performance targets; meanwhile, a smaller number is required to track non-maneuvering targets. The proposed fuzzy logic system showed good performance in tracking both maneuvering and non-maneuvering targets when applied to track-while-scan (TWS) radar.