Sequential Gaussian Approximation Filter for Target Tracking With Nonsynchronous Measurements

Abstract

This paper presents an adaptive sequential fusion estimation method for the target tracking with nonsynchronous measurements in wireless sensor networks (WSNs). Based on Gaussian assumption and Bayesian inference, a sequential cubature Kalman filtering (SCKF) method, as well as its square root form (SR-SCKF), is presented by applying the cubature rule to approximate the $function \times Gaussian$ integrals. By taking into consideration the time-varying properties of the measurement noise and the linearization errors, some adaptive factors are introduced into the SCKF to compensate for the measurement uncertainties based on Chi-square tests. The convergence analysis of the SCKF is presented. It is proved that the adaptive SCKF (ASCKF) has a better convergence property than the SCKF. Both simulations and experiments of a target tracking example are presented to show the effectiveness and superiority of the proposed ASCKF method.