Use of Centrality Metrics to Determine Connected Dominating Sets for Real-World Network Graphs

The high-level contribution of this paper is to demonstrate the use of centrality metrics to determine connected dominating sets (CDS) in large complex graphs representing a variety of real-world networks. We consider the following centrality metrics: Degree, Eigenvector, Betweeness and Closeness. We implement algorithms to determine each of these centrality metrics on large complex network graphs. We run these algorithms on six classical real-world networks (of size 34 to 332 nodes) that are commonly used in network analysis studies. We observe the Betweeness centrality to yield CDS of the smallest size (number of nodes and edges constituting the CDS) in five of the six networks. We observe the Degree-based centrality to return the lowest CDS size for three of the six networks (all tied with Betweeness centrality). We observe the Closeness centrality-based CDS to be of the smallest size for the smallest of the six networks, but incurs the largest size for the other five networks.