TeamSense: Energy-Efficient Autonomous Mobile Wireless Sensor Networks for Object Tracking

A typical surveillance application involves the use of limited sensing resources to monitor highly dynamic and uncertain environments. As a result, there is increasing demand to develop self-optimizing wireless sensor networks that are capable of adapting to such conditions. In this paper, we present a modeling framework, TeamSense, for designing and evaluating the performance of autonomous mobile wireless sensor networks for object-tracking applications. We program each sensor node to allow peer communication and collaboration to track several objects concurrently while reducing the network's overall energy consumption. Operating in dynamic, decentralized teams, sensor nodes continuously fuse gathered data to enhance the estimation and prediction of the objects' trajectories and optimally reconfigure the formed teams. The optimal reconfiguration of each team is obtained through solving a mixed-integer program. We perform a set of experiments to examine the performance of the developed network. Our results show that the TeamSense algorithms are computationally efficient and achieve a high level of tracking performance while using limited sensing resources and maintaining desirable energy consumption levels.