Noise‐aware manoeuvring target tracking algorithm in wireless sensor networks by a novel adaptive cubature Kalman filter

Abstract

Current target tracking algorithms for wireless sensor networks in noise environments have large positioning errors. Owing to the environmental noise, Kalman filters (KFs) are used to estimate the target position. To reduce the adverse effect of unknown or time-varying noise on KFs, adaptive KFs (AKFs) are developed. However, the present AKFs can only achieve second-order estimation accuracy. To improve the existing target tracking algorithm's positioning accuracy under unknown and time-varying noise environments, the authors propose a noise-aware algorithm based on a novel third-order adaptive cubature KF (ACKF) with higher estimation accuracy, which improves the accuracy of the existing algorithm by up to 63%. The innovative ACKF contains a new third-order noise statistic estimator and a traditional cubature KF without noise perception. A large number of numerical simulations and practical experiments show that the proposed noise-aware target tracking algorithm based on the novel ACKF is always more accurate than the target tracking algorithms based on the current KFs, no matter whether the moving target is manoeuvring or not, whether the strength of the noise is small or large, whether the number of anchor nodes is many or few, and whether the noise is time-varying or constant.