Analysis and Design of Protocols for Clustering in Mobile Ad Hoc Networks

Abstract

Communication in mobile ad hoc networks (MANET) without having any fixed infrastructure has drawn much attention for research. The infrastructure based cellular architecture sets up base stations to support the node mobility. Thus, mapping the concepts of base stations into MANET could meet its challenges like limited battery power, scalability, available band width etc..This leads to the design of logical clusters, where the cluster heads in every cluster play the role of base station. The cluster heads also form the virtual back bone for routing the packets in the network. In this thesis, simulation based survey has been made to study the strengths and weaknesses of existing algorithms that motivated for the design of energy efficient clustering in MANET. Neighbour Detection Protocol (NDP) has been designed to help the nodes to probe their immediate neighbours. In this protocol, every node broadcasts its own information to the network, so that it is received by a node that lies within its transmission range. The receiver senses its neighbours and updates its neighbour table from time to time. This protocol is validated through simulation by using Colour Petri Nets (CPN) prior to its implementation. Topology Adaptive Clustering Algorithm (TACA) has been proposed, that uses the node mobility and its available battery power for calculating the node weights. A node having the highest weight among its immediate neighbours declares itself as the volunteeer cluster head. As the current head consumes its battery power beyond a threshold, non-volunteer cluster heads are selected locally. The algorithm aims to utilise the battery power in a fairly distributed manner so that the total network life time is enhanced with reduced cluster maintenance overhead. During the process of clustering, some isolated heads without having any members are formed. This increases the delay in communication as the number of hops in the routing back bone is increased. A ransmissiion Range Adjustment Protocol (TRAP) has been proposed, that allows the isolated nodes to adjust their ranges to remain connected with existing cluster heads. The results show that, TRAP reduces the delay in communication by reducing the number of cluster heads in the network. Validation for the base protocol NDP and algorithm TACA are made through simulation by using the CPN tools. Each of the proposed work is evaluated separately to analyse their performances and compared with the competent results.