Probability-Based Prediction and Sleep Scheduling for Energy-Efficient Target Tracking in Sensor Networks
Target tracking is an important application in Wireless Sensor Networks or Internet of Things. In most applications, a large number of inexpensive and stationary sensor nodes are deployed randomly in a field to cooperatively monitor the intrusive targets that are able to move around the field. To save energy depletion, sensor nodes are in sleep mode in most of the time while they should be waked up. When a moving target is in their proximity. It needs at least three sensor nodes cooperatively work together to precisely locate a target. However, too many awake sensor nodes will not help the localization and tracking process but only waste sensor node's energy. Thus, in this paper, a cluster-based tracking algorithm is proposed. In this algorithm, a trajectory prediction scheme is designed to forecast the positions of the targets of interest as a probability measure, based on which suitable sensor nodes are waked up to keep on tracking and locating the targets of interest. The duty cycle, namely the sampling interval, is also decided to ensure the successful tracking rate. Our extensive simulation results show that our algorithm PCF outperforms the representative multi-target tracking algorithm, DMMT, in the literature.