Distributed measurement selection for energy-efficient radio tracking

We address the problem of energy reduction for a wireless sensor network performing an estimation task and propose an a posteriori measurement selection solution. The energy is saved by selecting in a distributed way a subset of the available measurements for transmission to a central processor. We show how to perform measurement selection for a network of radio-receivers tracking a target via the received signal strength indicator. The proposed measurement selection method is independent from the current estimate and based on the assumption that the best measurements have the largest received signal strength. The method operates through distributed max-consensus and schedules transmissions of the measurements without any centralized control. We compare the accuracy of our method with a centralized selection scheme and a Nearest Neighbor criterion. We show that the proposed measurement selection method consumes less energy for communication than centralized and distributed data fusion approaches, under the assumptions that radio nodes are always switched on to detect targets and that high-power long-range communication holds as a lower bound for the energy spent by the network for transferring data to the central processor.