Social Network Analysis Facilitates Cognition in Large Wireless Networks: Clustering Coefficient Aided Geographical Routing

Abstract

As large wireless networks for Internet of Things and cyber-physical systems emerge, they face challenges with new technology, scalability, and consequently effective cognition to execute networking functions. Beyond conventional sensing, researchers recently note the potential of social network analysis and inference to comprehend the networking status, but straightforwardly focus on the first-order statistics such as connectivity, degree, and centrality. Instead, we reveal a new alternative to equip network nodes with topological cognition by estimating the clustering coefficient. We further exploit such new cognitive capability to enable more efficient routing algorithms. Starting from random geometric graph to treat the communication range of a node as a disc, which is equivalent to examining the outage probability of links under stochastic geometry analysis of interference, the impacts of density and traffic of nodes on the connectivity can be analytically understood. Similar to the structural holes in social networks, the ad hoc networks suffer from the void region problem in the geographical routing, which is even more harmful in dense environments. Discarding the global routing table that demands non-scalable communication overhead, the clustering coefficient estimation of nodes as a new cognitive functionality can significantly alleviate the void problem to result in more effective ad hoc networking.