Publication

Abstract : In large wireless multi-hop networks, routing is a main issue as they include many nodes that span over relatively a large area. In such a scenario, finding smallest set of dominant nodes for forwarding packets would be a good approach for better communication. Connected dominating set (CDS) computation is one of the method to find important nodes in the network. As CDS computation is an NP problem, several approximation algorithms are available but these algorithms have high message complexity. This paper discusses the design and implementation of a distributed algorithm to compute connected dominating sets in a wireless network with the help of network spectral properties. Based on local neighborhood, each node in the network finds its ego centric network. To identify dominant nodes, it uses bridge centrality value of ego centric network. A distributed algorithm is proposed to find nodes to connect dominant nodes which approximates CDS. The algorithm has been applied on networks with different network sizes and varying edge probability distributions. The algorithm outputs 40% important nodes in the network to form back haul communication links with an approximation ratio ≤ 0.04 * ∂ + 1, where ∂ is the maximum node degree. The results confirm that the algorithm contributes to a better performance with reduced message complexity.