Non-parametric mobile node localization for IOT by variational Bayesian approximations adaptive Kalman filter

With the rapid development of wireless communication, wireless sensor network (WSN) has attracted considerable attention. Location information of wireless sensor network is an important application in indoor complicate environment. In line of sight (LOS) environment, accuracy of localization is very high. However, the accuracy of localization is highly degraded in indoor where the measurement may be contaminated by nonline of sight (NLOS) propagation. The NLOS error can bring a big effect. In this paper, we propose a method to alleviate the influence of the NLOS when NLOS measurement noise parameter is unknown. The algorithm identify the propagation condition between the anchor nodes (ANs) and mobile nodes (MN) firstly. After that, we adopt the Kalman filter (KF) for LOS measurement filtering. In NLOS environment, modified variational Bayesian approximation adaptive Kalman filter (MVB-AKF) is proposed to estimate the mean and measurement noise covariance to eliminate the influence of NLOS. The proposed method does not assume any statistical knowledge of the NLOS error. The efficacy of the proposed approach is demonstrated through the numerical simulation and experiment.