Mining and visualising wireless sensor network data

Abstract

In this paper, we propose a new network management framework for large-scale randomly-deployed sensor networks, called Energy Map, which explores the inherent relationships between the energy consumption and the sensor operation. Through nonlinear manifold learning algorithms, we are able to: 1) visualise the residual energy level of each sensor in a largescale network 2) infer the sensor locations and the current network topology through mining the collected residual energy data in a randomly-deployed sensor network 3) explore the inherent relation between sensor operation and energy consumption to find the dynamic patterns from a large volume of sensor network data for further network design, such as which set of sensors in a network will be the best candidates to be the future cluster heads, which is usually very important to develop a good sensor network protocol stack such as clustering algorithms and routing protocols.