Towards swarm intelligence for efficient routing in content centric opportunistic networks

Content distribution and retrieval in opportunistic networks is a promising paradigm due to its large spectrum of potential applications. In particular, routing in this paradigm needs efficient solutions due to the numerous challenges raised by opportunistic networks. In this paper, we introduce and demonstrate the potentiality of the swarm intelligence approach applied for this difficult communication context. First, we formalize the notion of optimal spatio-temporal path in opportunistic networks and propose an algorithm based on the adjacency matrix to compute the length of such path. Based on this definition, we introduce a simple swarm routing protocol which allows nodes in an opportunistic network that use a publish/subscribe communication paradigm to find the optimal path to route content towards subscribers. The protocol works in a decentralized way in which each node does not have any knowledge about the global topology. Via opportunistic contacts nodes update a dynamic scalar value (i.e. an utility function) which synthesizes their spatio-temporal proximity from subscribers of a given content (i.e. how close the node is in space and time to a subscriber of this content). This individual behavior applied by each node leads collectively to the formation of gradient fields between content users and content providers. Therefore, content routing simply sums up to follow the lowest gradient slope along this field to reach the users who are located at the minimum of the field. Via simulations, we demonstrate the existence and relevance of such gradient field and show that in highly sparse and localized networks the spatio-temporal gradient field can stabilize and helps to improve routing performance and resources compared to classical diffusion schemes. This proposition is a first brick towards more sophisticated and high performing swarm optimization techniques applied for opportunistic networks.