Multi-step information-directed sensor querying in distributed sensor networks

Sensor tasking is essential to many sensing applications in resourceconstrained wireless ad hoc sensor networks. In this paper, we present a multi-step lookahead algorithm for sensor selection and information routing. The algorithm is based on the informationdriven sensor querying (IDSQ) that uses mutual information as a utility meature for potential information contribution of individual sensors, and extend it to prediction of information gain over a finite horizon while balancing cost such as the number of communication hops. Simulation results on target tracking problems have shown that the multi-step lookahead algorithm significantly improves the tracking performance compared to the original greedy algorithm, when “sensor holes” are present in a sensor network.