Investigation of the effects of network density on the optimal number of clusters in hierarchical Wireless Sensor Networks (WSNs)

Abstract

Wireless sensors are promising and revolutionary technology. The sensor devices are typically deployed in large number to form a communication network for monitoring the physical environment. These devices are battery powered and the lifetime of the device is dependent on the battery's life. Once the battery's energy is depleted, the node is considered dead and is no longer part of the network. Recharging the sensors' batteries is not feasible, and in some cases, is completely impossible. Due to their severe energy constraints and redundancy in sensed data, hierarchal architectures are usually suggested for these networks. These architectures have proven effective in extending the network life several folds. The LEACH network routing protocol is the first protocol to use clustering for energy conservation in Wireless Sensor Networks. LEACH designers observe that there is an ideal percentage of nodes that need to be cluster heads to achieve optimal energy performance. For their work and performance analysis they selected 5% as the number of cluster heads in the netwok. The majority of routing protocols for WSNs follow LEACH and use clustering for energy performance optimization. Most of these protocols have taken the 5% percentage of cluster head as their ideal working setting, without independently qualifying this assumption. In this work, we use simulation modeling to investigate the dependency of this cluster head percentage on the network node density. Our results show that this percentage is not universal for all network settings, and is indeed dependent on the density. These findings will add to the challenging task of WSNs hierarchical networking protocols design.