An energy-efficient prediction-based algorithm for object tracking in sensor networks

Object Tracking Sensor Network (OTSN) is considered one of the most energy consuming applications of wireless sensor network. OTSN is used to track moving objects and report their newest location which consumes a large amount of energy. However, energy of sensor node is limited and the movement of objects generally follows some definite patterns. We can reduce the energy consuming by predicting the next location of an object to keep irrelevant sensor nodes sleepy as long as possible. In this paper, we propose an energy-efficient prediction-based tracking algorithm called Improved Mining Pattern (IMP). This algorithm predicts the next active sensor node based on the backward dependence. The predicted paths can be updated partly fast through clustering. Besides, IMP reduces the long distance communication between sensor nodes and the base station. In addition, missing objects can be tracked again quickly through recovery algorithm which is based on prediction results. Moreover, this algorithm can track multi-species simultaneously. Experimental results show that IMP behaves better than other algorithms in reducing the energy consumption and the missing rate.