DEC: Diversity-Based Energy Aware Clustering for Heterogeneous Sensor Networks

The clustering schemes devised so far mostly concentrate on satisfying the classic requirement of wireless sensor networks, which is energy awareness, to prolong network lifetime. On the other hand, there is a growing demand on applying heterogeneous nodes in sensor network applications, like distributed composite event detection that introduces new requirements such as maximizing sensor diversity for each cluster. The devised clustering schemes usually ignore this important feature. In this paper we introduce a diversity-based energy aware clustering protocol (DEC) that not only uses the energy level of the nodes in electing cluster heads, but also produces clusters with higher diversity of sensor types in their members. We will use the term complete clusters for the clusters which have the maximum diversity of sensor types. Simulation results support the idea that DEC partitions the network with higher percentage of complete clusters which can be up to 40% more than the generated clusters by ACE and CPCP algorithms in a network with the connectivity degree of 10. Besides, cluster heads in DEC have higher energy level compared to aforementioned clustering schemes.