On the generation of stable communities of users for dynamic mobile ad hoc social networks

We present the principles, implementation and evaluation of the SAw-SHARC algorithm, a Stability Aware Sharper Heuristic for Assignment of Robust Communities. The objective of this contribution is to perform distributed detection of densely and reliably interconnected users of a wireless ad hoc network and group them in communities. Community assignment is achieved using the computation of a neighborhood similarity measure in conjunction with a node-centric estimation of the communication links relative quality and epidemic propagation of community labels. Our algorithm also implements counter-measures to prevent domination of a single community in highly dynamic networks. We consider this contribution as a valuable tool for the development of large ad hoc mobile social networks (MoSoNets) where person-to-person interactive applications require a high quality of service in communications. Identifying subsets of reliably interconnected users would allow to meet those requirements. Besides, it will help cope the scalability of those networks, while maintaining their user-oriented nature. We evaluated our contribution using simulation of meaningful scenarios of social interactions and ensured its performance. Results show that SAw-SHARC favors creation of size-bounded stable communities and prevents the community domination effect in dynamic networks.