Data Fusion Algorithms with State Delay and Missing Measurements

In wireless sensor networks and other engineering systems, there are situations wherein some delays occur in data transmission and some measurements might be randomly missing. This would cause inaccuracies in Kalman filter or its equivalent algorithms, when used for target tracking. In this paper four alternative algorithms are studied and the modifications to include the state delay and randomly missing measurements are provided. Especially:i) the gain fusion, H-infinity a posteriori, H-infinity risk sensitive filter, and H-infinity global filtering algorithms are modified, and evaluated for sensor data fusion scenario using numerical simulations carried out in MATLAB; and ii) a nonlinear observer based on the continuous time data fusion filter is presented, and asymptotic convergence result is derived using Lyapunov energy functional; these two aspects are the novel contribution of this paper.