ANALYSIS OF SENSOR FUSION USING KALMAN FILTER IN HOMOGENEOUS AND HETEROGENEOUS SWARM NETWORKS

Abstract

A swarm consists of simplistic mobile nodes sharing information to achieve a common complex task. Most real world sensors result in noisy data which must be corrected to achieve accurate results. Noise filter or estimation algorithms are very useful in achieving this task. Sensor fusion is another useful technique to attain more useful and meaningful data from multiple sources of less accurate data. In the proposed system a group of mobile nodes are used to measure a changing parameter of another underlying system. But the measurement taken by these nodes is assumed to be erroneous. Thus, an instance of the standard discrete Kalman filter is used at each node to obtain a more accurate estimate. In this dissertation, two sharing algorithms are proposed for homogeneous and heterogeneous swarms. Each algorithm uses two different types of sharing schemes, complex and simple. A simple analysis of the standard Kalman filter is conducted to observe its characteristics. The comparative analysis of the two proposed algorithms and a standard moving average shows improved performance of both algorithms. Also, the behaviour of the algorithms is studied with change in number of member nodes, sensor range, communication range, sensor accuracy and type of sharing in the system. It is also shown that the heterogeneous swarm performs better with the proposed heterogeneous algorithm than the homogeneous algorithm which does not account for the difference in each node.