Geographic Multicast with K-Means Clustering for Wireless Sensor Networks

Geographic routing has been proven to be efficient to provide scalable unicast routing in resource-constrained sensor networks. However, its applications in multicast routing remain largely unexplored. Recently GMR (Geographic Multicast Routing) and DCGM (Destination Clustering Geographic Multicast) have been proposed, which preserve the distributed computation of geographic routing while delivering data packets to multiple destinations with efficient routes. To further reduce the number of transmissions, we apply a clustering strategy to GMR and DCGM. The strategy improves the performance of GMR and DCGM by dividing the destinations into many clusters and sending the packet first to the closest destination in each cluster, which then sends the packet to other nodes in the cluster. Simulation results show that the strategy can reduce the number of transmissions up to 35% percent.