Object Tracking by mining movement Trajectories in Wireless Sensor Networks

Most of the recent research on Object Tracking Sensor Networks has focused on collecting all data from the entire sensor network and placing it into the sink, which delivers the predicted locations to the corresponding nodes in order to predict an object's movement. This collection method affects the freshness of the data and creates latency in predicting movement patterns. In addition, due to the great amount of packets being sent and received, the sensor nodes' energy is quickly exhausted. Although this data collection method might result in a higher accuracy rate for prediction, it does not extend the lifetime of the sensor network. In this paper, a distributed method is proposed in using the network structure of convex polygons. These polygons are cooperated to find the trajectories of an object and then these trajectories are used to predict objects' movement. The proposed method, based on Trajectory-tree Construction, should reduce both the storage space of collected trajectories and the time spent on trajectory prediction analysis. Simulations show that the proposed method can reduce the energy consumption of the nodes and can extend the lifetime of the network in efficient.