Design of an Augmented Clustering Model using LEACH & PEGASIS for Low Complexity & High Network Lifetime

Abstract

Low energy clustering models are heart of wireless networks, which assist in reducing node-level power consumption via application of geometric identities. These models allow network designers to use the same configuration wireless network for a longer duration, thereby assisting in better network lifetime. A wide variety of system models are proposed for this, and it is observed that Low-energy adaptive clustering hierarchy (LEACH), and Power-Efficient Gathering in Sensor Information Systems (PEGASIS) provide lightweight and high efficiency clustering for wireless networks. But, both of them have their own shortcomings, which limits their real-time scalability. For instance, LEACH forms random & variable sized clusters which are distributed randomly and unevenly, and it uses node-to-cluster head (CH) communications even in cases where single hop communications are possible. While, PEGASIS, uses Greedy Algorithm for chain formation, which cannot find globally optimal cluster head, and it also assumes that all nodes have knowledge about other nodes in the network, due to which it cannot be applied to adhoc network scenarios. To remove these drawbacks from both the models, this paper proposes design of a novel augmented clustering method that combines advantage of both LEACH & PEGASIS (ACPL) to improve model scalability, while maintaining better energy efficiency. The model initially uses the node-discovery capabilities of LEACH for stochastic cluster head selection, and once all nodes are discovered, then PEGASIS is applied for chain formations. These chains are used for cluster-to-cluster communications, and intelligently decide when to directly communicate between node-to-base station depending upon distance & residual energy values. The combined model was tested on a wide variety of network deployments, and it was observed that the proposed was able to reduce energy consumption by 16% when compared with LEACH, 8.6% when compared with PEGASIS, and 9.3% when compared with other state-of-the-art clustering models. The proposed model was also observed to be faster than existing models due to use of direct node-to-base station communications, which assisted in improving its throughput by 3% when compared with LEACH, 1.9% when compared with PEGASIS, and 5.4% when compared with other state-of-the-art approaches. The model was also tested on different network sizes, and its performance was observed to be consistent across them, thereby making them scalable and usable for a wide variety of network deployments.