Distributed estimation of statistical correlation measures for spatial inference in WSNs

This work shows how to obtain distributively important statistical measures such as the semivariogram and the covariogram in a Wireless Sensor Network. These statistics describe the spatial dependence of the sensed area and allow making inferences about unknown field data. In practice, these are complex measures that require global knowledge such as the distance between every pair of nodes, which is not available in a distributed scenario. Then, motivated by the distributed nature of a Wireless Sensor Network and the requirement of making estimations in many real applications, we propose a distributed method to obtain an approximation of these measures, based only on the local samples of the nodes. Our method only requires knowing, at each node, the geographic position of its neighbors. Additionally, we show that introducing random movements of the nodes, the quality of the results can be improved. Simulation results are presented to evaluate the performance of our algorithm.