Abstract: By analyzing the cluster heads selection unreasonable and loading unbalancing of the network in Wireless Sensor Networks（WSNs）, this paper presents a novel energy balanced multistage clustering algorithm, which cluster head makes superior decision by analyzing learned data from surroundings network. In the cluster head election phase, overall considering residual energy and the relative node density, it can choose suitable cluster heads; In the routing phase, choose the cost-
effective communication solution in the cluster by using greedy algorithm, meanwhile, preserve some energy for the data forwarding between the clusters; Cluster head can detect some death nodes and broadcast the death message, more effective to maintain the network running. Simulation results show that it can choose more reasonable cluster head, balance the energy load of all nodes effectively and significantly prolong the network lifetime.

Key words: wireless sensor networks, energy balanced, multistage clustering, node density, network lifetime

摘要： 通过分析无线传感器网络分簇路由协议中簇首选择不合理和网络负载不均衡的问题，提出一种能耗均衡的多级分簇算法，簇首通过分析已学习到的周围网络数据作出较优决策。在簇首选举方面，综合考虑了节点剩余能量和相对节点密度，选择出合适的簇首；路由方面，运用贪婪算法选择较优簇内通信方案，为簇间数据转发预留能量；簇首对死亡节点能及时发现和广播死亡信息，更好地维护网络运行。仿真和分析结果表明，该算法能选出更为合理的簇首，更有效地均衡了网络负载，显著延长了网络寿命。

关键词: 无线传感器网络, 能量均衡, 多级分簇, 节点密度, 网络寿命