MULTI-SENSOR DATA FUSION FOR TARGET CLASSIFICATION AND TRACKING

Abstract

Multi-Sensor Data Fusion in Wireless Sensor Networks is a fast emerging field with very diverse applications in the civilian and military domains. Being a relatively fresh area there is much scope for research to include applications like environmental monitoring, industrial sensing and diagnostics, infrastructure protection, battlefield awareness etc. to name a few. A potential research domain is in battlefield awareness wherein an important application is target tracking. Cross border infiltrations by hostile forces is one such scenario where a tactical deployment of wireless nodes employing suitable tracking algorithms can be used to check such infiltrations. One of the most fundamental and widely used approaches to target tracking is the Kalman filter. It is well known that in presence of unknown noise statistics there are difficulties in the Kalman filter for yielding good results. Tuning of the filter statistics is an important research area. It is observed that generally in Kalman filter operation, the gain tends to a steady state value after the initial transients. Hence working directly with constant Kalman gains, it is possible to obtain good tracking results dispensing with the use of the usual covariance. The thesis is aimed at addressing the specific issues of target tracking by developing a suitable algorithm. The three types of filters studied are the reference Kalman filter, tuned Kalman filter and constant gain Kalman filter for both the linear and nonlinear state variable models. Our numerical studies show that the constant gain Kalman filter gives good comparative performance in both the standalone and multi-sensor data fusion modes for the target tracking problem. Our significant finding is that the constant gain Kalman filter circumvents or in other words trades the gains with the filter statistics which are more difficult to obtain. The study has been carried out for both the linear and nonlinear state variable models in Multi-Sensor Data Fusion scenario which we can term as "Multi Sensor Fusion Approximation based Tuned and Constant Gain Kalman Filter Approach for Target Tracking".