为了解决传感器通过三层网络结构向融合中心传递数据时必须进行分簇的问题,提出了一种采用图论方法进行聚类分簇的具体实现算法及实现过程,并通过实验验证了图论分布式聚类算法相对于集中式K-平均算法的性能优越性。
A concrete algorithm for cluster based on graph theory and the realization process is proposed in order to improve the sensor′s cluster performance of triple-network in transmitting data to the aggregation center of the network. Experiments show the distributed algorithm based on graph theory has advantage in performance over the serial K-means algorithm.