A Research On K-means Clustering Algorithm Based On Feedback

Clusters in the classical LEACH routing protocol are not uniform and cluster heads are random in wireless sensor network, so we proposed a K-means clustering algorithm based on feedback of the number of dead nodes. In the paper,K-means-based clustering was used for cluster formation and the center node of the cluster was selected as the cluster head. We got the several nodes as super cluster heads based on location centrality . Among them,the biggest one was chosen as SCH. Later on,We only used adaptive scoring function controlled by the number of dead nodes to select cluster heads. The function of super-cluster heads was just to aggregate information and transfer it. Simulation results show that compared to LEACH and traditional method based on K-means, the proposed algorithm improved the lifetime of network by 35% and 25% respectively. It was also proved that feedback mechanism and super-cluster heads were beneficial to the improvement of the network lifetime.