Clustering Overhead and Convergence Time Analysis of the Mobility-based Multi-Hop Clustering Algorithm for Mobile Ad Hoc Networks

Abstract

With the emergence of large mobile ad hoc networks, the ability of existing routing protocols to scale well and function satisfactorily comes into question. Clustering has been proposed as a means to divide large networks into groups of suitably smaller sizes such that prevailing MANET routing protocols can be applied. However, the benefits of clustering come at a cost. Clusters take time to form and the clustering algorithms also introduce additional control messages that contend with data traffic for the wireless bandwidth. In this paper, we aim to analyze a distributed multi-hop clustering algorithm, Mobilitybased D-Hop(MobDHop), based on two key clustering performance metrics and compare it with other popular clustering algorithms used in MANETs. We show that the overhead incurred by multi-hop clustering has a similar asymptotic bound as 1-hop clustering while being able to reap the benefits of multi-hop clusters.

Mitigating the impact of node mobility on ad hoc clustering

“…Mobile agents also help in cluster size adjustment, re-clustering, and continuous cluster state monitoring. In other solutions [19][20][21][22], the mobility pattern of the nodes is considered for forming the clusters. In Reference [21], the mobile nodes are organized in variable-sized non-overlapping clusters, while in References [19,20] the clusters are of variable diameter.…”

“…In other solutions [19][20][21][22], the mobility pattern of the nodes is considered for forming the clusters. In Reference [21], the mobile nodes are organized in variable-sized non-overlapping clusters, while in References [19,20] the clusters are of variable diameter. The algorithm proposed in Reference [21] uses a combination of both physical and logical partitions of the network for the clustering, while similar moving nodes are grouped together in References [19,20].…”

“…In Reference [21], the mobile nodes are organized in variable-sized non-overlapping clusters, while in References [19,20] the clusters are of variable diameter. The algorithm proposed in Reference [21] uses a combination of both physical and logical partitions of the network for the clustering, while similar moving nodes are grouped together in References [19,20]. Basu et al [22], similarly to the lowest ID solution, propose a weight-based clustering algorithm where aggregated local mobility (ALM) is used to elect a clusterhead.…”