Community detection in opportunistic networks using memory-based cognitive heuristics

In a pervasive networking scenario like the Cyber-Physical World convergence, personal mobile devices must assist their users in analysing data available in both the physical and the virtual world, to help them discovering the features of the environment where they move. Mobile Social Networking applications are an example of Cyber-Physical applications, supporting users in their interactions in both worlds (e.g., during physical encounters, as well as during online interactions). It is very important, therefore, that nodes autonomously detect latent and dynamically changing social structures, resulting from common mobility patterns of users and physical co-location events. To this end, in this paper we propose a novel dynamic and decentralised community detection approach, whereby the nodes´ behaviour is inspired by that of their human users, if they were exposed to the about physical encounters with other users, and would have to perform the same detection task. Specifically, we use cognitive heuristics, which are simple, low resource-demanding, yet effective, models of the human brain cognitive processes. At each node, the approach proposed in this paper, starting from the observed contact patterns with other nodes, estimates the strength of social relationships and detects social communities accordingly. An initial simulation evaluation shows that nodes are able to correctly identify the social communities that exist in their environment and to efficiently track change of membership due to modifications of the users´ movement patterns.