Cognitive structure adaptive particle filter for radar manoeuvring target tracking

Abstract

The accuracy and efficiency of the particle filter algorithm partly depend on the number of particles used in the estimation. The number of particles is specified beforehand and kept fixed in the process of radar manoeuvring target tracking (MTT). In practice this may be highly inappropriate since it ignores the uncertainty in the models and the varying dynamics of the processes. A cognitive structure adaptive particle filter (CSAPF) algorithm is proposed to increase the accuracy and efficiency by adapting the number of particles online. The name CSAPF is due to the fact that the authors change the number of particles dynamically with radar's real-time perceptual information entropy for MTT. The authors cognitive structure approach chooses a transmitted waveform to improve tracking accuracy and chooses the lower bound of required particles to improve tracking efficiency. Monte–Carlo simulation results show that the approach has significant improvements over the particle filter with a fixed number of particles and over the fixed waveform particle filter algorithm.