Multisensor Data Fusion with Estimated Weights

The desired improvements of a tracker rely on more accurate state estimates and less computation loads. A state-vector multisensor data fusion approach that consists of local processor and global processor is described for the problem of tracking a maneuvering target in the local inertial cartesian coordinate system (LICCS). For local processor, the sensor-based filtering algorithm utilized in the reference cartesian coordinate system (RCCS) is presented for target tracking when the radar measures range, azimuth and elevation angles in the spherical coordinate system (SCS). To reduce the computational loads involved in physical implementation, the decoupling technique that Kalman filter gain formulations are recursively computed in the line-of-sight cartesian coordinate system (LCCS) and then transformed for use in the RCCS is adopted. For global processor, the data fusion algorithm, called covariance matching method, is devised using sensor filter covariance matrices to estimate each sensor weight for obtaining an improved joint state estimate in the LICCS. Simulation results are presented comparing the performance of the proposed algorithm with the sensor-based decoupled Kalman filtering algorithms. That the number of sensors may influence the tracking accuracies of proposed algorithm is discussed