Correlation-Model-Based Data Aggregation in wireless sensor networks

Environment monitoring is the typical application of the wireless sensor networks. And for physical systems, there is a widespread principle that the difference between temperatures of adjacent nodes is an approximate constant. In this paper, we propose a constant correlation model and a Correlation-Model-Based Data Aggregation algorithm (CMB) according to this principle. CMB mines the correlation model of every pair of adjacent nodes, and draws a correlation graph of the network, then builds an aggregation routing tree for each connected component of correlation graph by following the shortest path algorithm. Following the CMB algorithm, the data of the child nodes will be suppressed, once the data of a certain node and that of its children´s meet the restriction. When the aggregated data arrive at the Sink node, all the data can be recovered. The recovery error will be limited within a specified small error threshold based on the reversed mechanism. The simulation based on real temperature data show that CMB algorithm has excellent performance in aggregation degree and average error.