Performance Evaluation of State Estimators for Airborne Target Tracking Using Multi Sensor Data Fusion

The main function of range sensory systems under a given dynamic environment is to detect, discriminate and track a particular target for surveillance in case of a friendly target or an enemy target interception. The combination of two or more sensors will provide better position estimate than a single sensor. In this paper, the advantages of the multi sensor data fusion is presented and compared over conventional single sensor tracking. The state estimation techniques are utilized to enhance position accuracy in a single and multi-sensor environment. The performance of each state estimator is evaluated by considering different target motions along with their nonlinear characteristics. The state estimators presented here varies from simple linear filters such as fixed gain and Kalman filters to complex nonlinear filters such as Particle filter. Two widely used Extended Kalman filter based fusion architectures such as measurement fusion and state vector fusion are explored. The data is simulated from two ground based sensors RADAR and FLIR (forward looking infra red) to examine the fusion process. The RMS error is computed in range, azimuth, and elevation angles. A complete mathematical modeling and simulation is implemented in MATLAB. It is found that fusion architectures have demonstrated better performance in tracking accuracy.