An Object Tracking Technique in Wireless Sensor Network Based on Prediction

Motion features have not been considered synthetically in all kinds of present object tracking techniques. A new object tracking techniques based on prediction is proposed. At firstit decides motion features of the object depending on current measurements or history records of its motion, and then predicts its future location by integrating with its current information such as location, velocity, and motion direction. When prediction fails, the network will start hierarchically recovery process according to its records on motion history or a priori knowledge. The simulation demonstrates that the approach reduces the number of sensor nodes waken up, thus lower energy consumption of nodes so as to prolong the longevity of sensor networks.