Title: Adaptive tracking of group targets using random matrix

Authors: Yun Wang; Guo-Ping Hu; Peng Ni; Lin-Hai Gan

Addresses: School of Air and Missile Defense, Air Force Engineering University, Xi'an, China ' School of Air and Missile Defense, Air Force Engineering University, Xi'an, China ' Unit 94921, Jinjiang, China ' School of Air and Missile Defense, Air Force Engineering University, Xi'an, China

Abstract: Adaptive tracking method of group targets using random matrix based on the current statistical model (CS) is presented in order to improve tracking performance of group targets in strong manoeuvring and high measurement error. In the estimation of group centroid, a bell-shape function is utilised as fuzzy membership function to adjust the maximum acceleration which can modify the process noise variance adaptively. The introduction of strong tracking filter with multiple suboptimal fading factors adjusts error covariance of predicted group centroid state adaptively when group targets manoeuvre strongly. In the estimation of group extension, the measurement accuracy is considered in the extended state estimation where the likelihood function is formatted with extended state and measurement error covariance which is calculated by the innovation and adaptive fading memory updated in iterative process. The simulation results show that the method presented in the paper achieves better tracking performance of group targets in strong manoeuvring and high measurement error compared with the existing methods.

Keywords: group targets; random matrix; current statistical modelling; bell-shape function; strong tracking; innovation covariance; adaptive tracking; tracking performance; fuzzy membership function; simulation.

DOI: 10.1504/IJWMC.2016.075220

International Journal of Wireless and Mobile Computing, 2016 Vol.10 No.1, pp.83 - 89

Received: 11 Jun 2015
Accepted: 01 Jul 2015
Published online: 08 Mar 2016 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article