A Communication-Efficient Distributed Clustering Algorithm for Sensor Networks

Sensor networks usually generate continuous stream of data over time. Clustering sensor data as a core task of mining sensor data plays an essential role in analytical applications of sensor networks. Although several algorithms have been proposed to address the problem of distributed clustering, in the domain of sensor networks these algorithms face major new challenges such as limited communication bandwidth and constraints in power supply, and storage resources. Moreover, previous studies about clustering in sensor networks have mostly focused on clustering sensor nodes and designing better network topology for the purpose of energy conservation rather than clustering sensor data for future analytical purposes. In this paper a communication efficient distributed algorithm is proposed for clustering sensory data. This approach addresses the limited bandwidth issue through summarized transmissions. Furthermore communication efficiency of the algorithm contributes to reduced power consumption. Time efficiency of the algorithm is evaluated through simulation experiments and the results are presented.