A dynamic clustering framework for target tracking in wireless sensor networks

In this paper we describe a dynamic clustering framework for target tracking. The clustering process is divided into three phases: a) prediction, the cluster head predicts where the target will be; b) new cluster head selection, if the prediction result implies that the target will move out of the cluster, then a new cluster needs to be formed, and firstly a new cluster head is selected; c) new cluster forming and taking over the tracking task. We study this dynamic clustering problem and give our solutions composed of a Taylor-series based prediction algorithm and other algorithms and steps for cluster head selection, which is based on the proportion of node´s residual energy and its initial energy, and the new cluster size´s determination which is based on some restrictions such as the scaping probability restriction and the energy restriction.