A data mining approach to energy efficiency in Wireless Sensor Networks

There has recently been a considerable amount of research work on using data compression techniques to minimize the volume of transmitted traffic, and consequently assist in reducing power consumption levels in Wireless Sensor Networks. In this paper, we present a data Oriented approach called Modelbased Clustering (MBC) which shrinks the communication flows between sensor nodes and sink node in a way that contributes to reducing power consumption in wireless sensor networks. The proposed work utilizes the capabilities of mixture-model based clustering to exploit both the temporal locality and slowly varying properties of the sensed data to model the sensor network's traffic. The generated models will be utilized by both the sensor nodes to process the sensed raw measurements and sink node to recover the original data without requesting those data to be completely transferred by the limited resources' sensor nodes. Results show that our approach contributes to decreasing energy consumption in resource-limited sensor network.