Selective forwarding for energy-efficient target tracking in sensor networks

• We apply a selective communication policy based on MDPs to target tracking.
• We set importance values to packets for identifying the relevance of the information.
• We show that selective policies complement data aggregation and fusion techniques.
• Selecting properly the sampling rate, lifetime prolongs without increasing the error.

Selective communication (censoring) strategies allow nodes in a sensor network to discard low importance messages with the purpose of saving energy that can be used for transmitting more important messages later. In this paper we apply simple selective policies based on Markov Decision Processes to a distributed target tracking scenario based on particle filters and the Information-Driven Sensor Querying (IDSQ) scheme. The resulting algorithm is combined with other energy-efficient schemes, such as data aggregation or data fusion, extending the action space of these techniques. Our simulation work shows that the network lifetime can be substantially increased while keeping a low tracking error. And even more, selecting the sampling rate properly, lifetime is prolonged without increasing the tracking error.