International Journal of Communication Systems
| Hi-index | 0.00 |
We consider the challenge of organizing densely deployed sensor nodes into the form of clusters, using the distribution of network residual energy (NRE), which is defined as the sum of node residual energy. Irrespective of network topology, the distribution of NRE is proven to approach Gaussian in dense node deployment. A decentralized clustering algorithm is present, using timers and a recursively updated probability to select nodes with more residual energy to become Cluster Head (CH) nodes and organize other nodes in the form of clusters over slotted time intervals. Embracing the dense node deployment, each node initializes its probability of becoming a CH node using the distribution of NRE defined in its neighborhood area. Each of the selected CH nodes resides in the center of its cluster area, which has a radius that can be arbitrarily chosen. The performances of the new clustering algorithm are analyzed and then validated via extensive simulations, taking into account variable cluster radius and variable network density. The new clustering algorithm significantly prolongs the network lifetime, in comparison to several representative and competing clustering algorithms reported in the literature. Copyright © 2010 John Wiley & Sons, Ltd. A decentralized clustering algorithm uses timers and a recursively updated probability to select nodes with more residual energy to become Cluster Head (CH) nodes and organize other nodes in the form of clusters over slotted time intervals. Each node initializes its probability of becoming a CH node using the distribution of Network Residual Energy (NRE) defined in its neighborhood area, which is proven to approximate the Gaussian distribution. The new clustering algorithm selects energy-rich CH nodes that reside in the center of every cluster area, and thus significantly prolongs the network lifetime. Copyright © 2010 John Wiley & Sons, Ltd.