A data correlation-based wireless sensor network clustering algorithm

Clustering is an energy efficient method for wireless sensor network monitoring. Existed algorithms are mostly location-based, where physical temporal-spatial correlations are considered, leading to a poor data compression rate. This paper proposes a novel data correlation based clustering approach. A greedy algorithm, minimum degree first clustering algorithm, is introduced to cluster the sensor network. To deal with correlation failures, periodic re-clustering and failure-sensitive delta clustering are proposed. The experimental results show the proposed algorithm can prolong the net work data compression rate of cluster head and the data restoration rate of base station. The network life time is prolonged as well.