Distributed Energy-Efficient Hierarchical Clustering for Wireless Sensor Networks Ping Ding, JoAnne Holliday, Aslihan Celik {pding, jholliday, acelik}@scu.edu Santa Clara University Abstract: Since nodes in a sensor network have limited energy, prolonging the network life- time and improving scalability become important. In this paper, we propose a distributed weight-based energy-efficient hierarchical clustering protocol (DWEHC). Each node first locates its neighbors (in its enclosure region), then calculates its weight which is based on its residual energy and distance to its neighbors. The largest weight node in a neighborhood may become a clusterhead. Neighboring nodes will then join the clusterhead hierarchy. The clus- tering process terminates in O(1) iterations, and does not depend on network topology or size. Simulations show that DWEHC clusters have good performance characteristics. 1 Introduction Sensor nodes are relatively inexpensive and low-power. They have less mobility and are more densely deployed than mobile ad-hoc networks. Since sensor nodes are always left unattended in sometimes hostile environments, it is difficult or impossible to re-charge them. Therefore, energy use is a key issue in designing sensor networks. Energy consumption in a sensor network can be due to either useful or wasteful work. Useful energy consumption results from transmitting/receiving data, querying requests, and forwarding data. Wasteful energy consumption is due to collisions and resulting retransmis- sions, idle listening to the channel, and overhead of each packet header (even when the data packet is short). Energy consumption reduces network lifetime, which is defined as the time elapsed until the first node (or a certain percentage of nodes [1]) use up their energy. To reduce energy consumption, clustering techniques have been suggested [3-20]. These techniques organize the nodes into clusters where some nodes work as clusterheads and collect the data from other nodes in the clusters. Then, the heads can consolidate the data and send it to the data center as a single packet, thus reducing the overhead from data packet headers. Clustering has advantages for: 1) reducing useful energy consumption by improving bandwidth utilization (i.e., reducing collisions caused by contention for the channel); 2) reducing wasteful energy consumption by reducing overhead. In a clustered network, the communication is divided into intra and inter cluster com- munication. The intra-cluster communication is from the nodes inside a cluster to the head. The inter-cluster communication is from the heads to the data center (sink node). The energy efficiency of a clustered sensor network depends on the selection of the heads. Hei- nzelman et al. [3] propose a low-energy adaptive clustering hierarchy (LEACH), which generates clusters based on the size of the sensor network. However, this approach needs a priori knowledge of the network topology. Younis and Fahmy [4] propose a Hybrid Energy-Efficient Distributed clustering (HEED), which creates distributed clusters without the size and density of the sensor network being known. However, the cluster topology fails to achieve minimum energy consumption in intra-cluster communication. Also, as we show in Section 5, the clusters generated by HEED are not well balanced. In this paper, our goal is to achieve better cluster size balance and obtain clusters such that each has the minimum energy topology. We propose a distributed weight-based