SENOCLU, Energy Efficient Approach for Unsupervised Node Clustering in Sensor Networks

Abstract

Acquisition and analysis of data from sensor networks, where nodes operate in unsupervised way, has become a ubiquitous issue. The biggest challenge in this process is related to limited energy, computational and memory capacity of sensor nodes. Therefore, the main goal of our work is to devise and evaluate the contribution of an energy efficient algorithm for data acquisition in sensor networks.

The proposed SENOCLU algorithm considers specific requirements of sensor network application like energy efficiency, state change detection, load balancing, high-dimensions of the sensed data etc. By applying these techniques, this algorithm contributes in filling the gap between distributed clustering and high-dimensional clustering algorithms that are available in the literature. This work evaluates the contribution of this algorithm in comparison to other competing state-of-the-art techniques.

The experiments show that by applying SENOCLU algorithm better life times of sensor networks are achieved and longer monitoring of different phenomena is provided.