Distributive Target Localization and Tracking in Sensor Networks

Abstract

Sensor networks are among emerging technologies that have ability to monitor and instrument the physical world. Sensors are inexpensive wireless devices (nodes) densely distributed over the region of interest and hence forming the sensor network. Nodes are connected to the base station via a command network, such as the Internet. Typically, each individual node can sense in multiple modalities but has limited communication and computation capabilities. In simple target tracking, data fusion techniques are mostly used where the sensors sense the environment and send their data to central node, a master node (powerful node), which tracks the target. But the possibility for failure at that single point is always there, if the central node (decision making node) fails this means the whole system fails. In this project, we laid emphasis on decentralized data fusion. There is a network of sensor nodes, each node having its own processing and communication facility where fusion occurs locally at each node on the basis of local observations and the information gathered from neighboring nodes. There is no common place where fusion or global decisions are made. This approach can be referred as a decentralized data fusion system. The tracking procedure is efficient to the extent to which power requirements (e.g. battery limitations, signal processing and communication overhead) allow. Simulation results clearly show that the track determined by our algorithm is very close to the actual track of the target which confirms the efficacy of the algorithm.