An energy efficient and resource preserving target tracking approach for wireless sensor networks

Efficient object tracking in wireless sensor networks (WSNs) is of importance in many application scenarios such as military and surveillance, health monitoring, etc. In this paper, we propose a target tracking technique from two approaches, dynamic clustering and predictive tracking techniques. We use the Markov Decision Process (MDP) to predict the position of the tracked object over time or the `state´. In addition, we devise a mechanism by dividing a cluster of a WSN into a set of mini-clusters which helps to reduce the number of active nodes at any given time and in turn reduce the energy consumption and data transmission during sensing. The experimental evaluations show the proposed approach can dynamically track and predict a moving object with reduced energy consumption and up to 40% less data generated.