DECADE: Distributed Emergent Cooperation through ADaptive Evolution in mobile ad hoc networks

Abstract

The scarce resources of a mobile ad hoc network (MANET) should not be wasted attending selfish nodes (those nodes that use resources from other nodes to send their own packets, without offering their own resources to forward other nodes’ packets). Thus, rational nodes (those nodes willing to cooperate if deemed worthy) must detect and isolate selfish nodes in order to cooperate only among themselves. To achieve this purpose, in this paper we present a new game theoretic trust model called DECADE (Distributed Emergent Cooperation through ADaptive Evolution). The design of DECADE is shown by first, analyzing a simple case of packet forwarding between two nodes, and then the results are extended to bigger networks. In DECADE, each node seeks individually to maximize its chance to deliver successfully their own packets, so that the cooperation among rational nodes and the isolation of selfish nodes appear as an emergent collective behavior. This behavior emerges as long as there is a highly dynamic interaction among nodes. So, for those cases where the mobility alone does not suffice to provide this interaction, DECADE includes a sociability parameter that encourages nodes to interact among them for faster learning and adaptability. Additionally, DECADE introduces very low overhead on computational and communication resources, achieving close to optimal cooperation levels among rational nodes and almost complete isolation of selfish nodes.