K-Centers Mean-shift Reverse Mean-shift clustering algorithm over heterogeneous wireless sensor networks

Abstract

A clustering algorithm K-centers mean-shift reverse mean-shift for heterogeneous wireless sensor networks is presented in this paper, addressing the empty cluster problem as a key issue. Many clustering algorithms for sensor networks have empty cluster problems due to random deployment, which causes resource and cost inefficiencies. Our algorithm calculates the mean-shift of sensor nodes and the reverse mean-shift of cluster heads to iteratively move cluster heads closer to the sensor nodes' density and away from cluster heads' density. This helps cluster heads better accommodate the distribution of sensors. Our proposed K-Centers Mean-shift Reverse Mean-shift algorithm decreases the number of empty clusters dramatically, and it also balances the sizes of clusters more evenly.