Hide/Show Apps

Dependability-based clustering in mobile ad-hoc networks

Ergenç, Doǧanalp
Eksert, Levent
Onur, Ertan
Clustering is an effective solution to handle challenges due to lack of centralized controllers and infrastructure in ad-hoc networks. Clustered network topologies offer better scalability and reliability when compared with flat topologies. In this paper, a novel clustering algorithm, Dependability-based Clustering Algorithm (DCA) is presented. In DCA, we focus on the dependability of clusters themselves rather than considering only individual nodes to maintain the clustered structure of a network. We validated the performance of our proposal by discrete-event simulations and showed that our approach outperforms its opponents in terms of stability, energy efficiency and quality of service in stationary and mobile scenarios. Moreover, we performed sensitivity analysis of different metrics that are used to select cluster heads and compute the cluster dependability score proposing a complete optimization framework to analyze such metrics for weighted clustering algorithms. As a result of the analysis, we found that while topology-related metrics have negative impact, in many cases, energy-related metrics (e.g., residual energy) have varying effects on different objectives