Generating Dominating Sets Using Locally-Defined Centrality Measures

Abstract

The dominating set problem has many practical applications but is well-known to be NP-hard. Therefore, there is a need for efficient approximation algorithms, especially in applications such as ad hoc wireless networks. Most distributed algorithms proposed in the literature assume that each node has knowledge of the network structure. We propose a distributed approximation algorithm that uses two rounds of communication, and where each node has only local information, both in terms of network structure and dominating set assignment. First, each node calculates a local centrality measure to determine whether it is part of the dominating set $D$. The second round guarantees $D$ is a dominating set by adding any non-dominated nodes. We compare several centrality measures and show that the Shapley value, introduced in game theory, is theoretically motivated and performs well in practice on several synthetic and real-world networks.