Maneuvering Target Tracking Using Current Statistical Model Based Adaptive UKF for Wireless Sensor Network

摘要

This paper presents Current statistical model based Adaptive Unscented Kalman Filter (CAUKF) for maneuvering target tracking, which is based on Received Signal Strength Indication (RSSI). In order to introduce the Kalman filter, the state-space model, which uses RSSI values as the measurement equation, needs to be obtained. Thus a current statistical model for maneuvering target based on the path loss model is presented. To avoid the negative influence of current statistical model's limited acceleration, the functional relation between the maneuvering status of target and the estimation of the neighboring position information is applied to carry out the adaptation of the process noise covariance Q(k). Then, a novel idea of modified Sage-Husa estimator is introduced to adapt the process noise covariance matrix Q(k), while the adaptive measurement noise covariance matrix R(k) is implemented by a fuzzy inference system. The experimental results show that the final improved CAUKF is an algorithm with faster response and better tracking accuracy especially in maneuvering target tracking.