Robust adaptive cubature Kalman filter for tracking manoeuvring target by wireless sensor network under noisy environment

Abstract

Abstract The existing adaptive Kalman filters for tracking manoeuvring targets by wireless sensor networks can easily lose robustness when both the measurement and process noises are unknown and time‐varying, resulting in large positioning errors. To solve this problem, a wireless sensor network manoeuvring target tracking algorithm using a novel robust adaptive cubature Kalman filter is proposed. This innovative robust adaptive cubature Kalman filter consists of a derived third‐order biased noise statistic estimator and conventional cubature Kalman filter. This derived noise statistic estimator can simultaneously sense time‐varying and unknown measurement and process noises and ensure that the adaptive cubature Kalman filter's robustness is not lost. The robustness of the novel robust adaptive cubature Kalman filter is strictly proven in this study. Extensive practical experiments and numerical simulations show that the proposed robust adaptive cubature Kalman filter always has higher target tracking accuracy than other existing adaptive Kalman filters, regardless of whether the mobile target is manoeuvring or not, the noise is unknown or time‐varying, and the number of anchor nodes is few or many.