A dynamic clustering framework for target tracking in wireless sensor networks
Bo Ouyang, Xinyu Jin, Jun Tang · 2010
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.