Weighted Degree-Based Host Clustering Algorithm in Grid

Haiyang Zhang, Huadong Ma

Abstract: Considering the computation and transmission ability of nodes, we propose a weighted degree-based clustering algorithm in grid. To achieve low clustering complexity, we have made use of the number of hops to constrain the range of selective nodes in grid. Although such solutions have resulted in a failure to obtain the optimal distances, under the case of continuous dynamical changes of nodes, the optimal distance is unnecessary. Our algorithm has applied weighted degree of nodes and the number of hops to select initial cluster centers. During the process of clustering, we have adjusted the cluster centers according to the maximum weighted degree of nodes to guarantee the data transmission ability of the nodes in every cluster. The simulation results demonstrate that our algorithm achieves better clustering performance than the K-Means heuristic algorithm and the density-based host clustering algorithm in three aspects: execution time, the number of effective clusters and average weighted capacity.