Intelligent cross-layer distributed data aggregation in wireless sensor networks

The task of energy efficient data aggregation in a wireless sensor network (WSN) is investigated. The challenge we focus on is to coordinate the decisions made distributively at the sensor nodes in a fashion to contribute to the common goal of computing the desired statistic while minimizing energy consumption. In this work, finding the maximum sensor readings in the WSN is considered as an illustrative example. Sensor nodes whose readings are smaller than a threshold and have yet to report their values will deactivate themselves and fall into the sleep mode to conserve energy. Remaining active sensor nodes will update the lower or upper bound of this interval and will re-estimate a threshold value in this interval that separates the maximum reading and the second largest reading. It is shown both analytically and empirically that the proposed cross-layer probabilistic data aggregation method offers superior performance while demanding much lower energy consumption than existing algorithms. In comparison to the naive method, which consumes power as O(Nlog(N)), our algorithm offers a marked improvement at O(1).