DASH: A Distributed Approach for Selfishness Handling in a DTN

Abstract

Delay and disruption tolerant network is deemed as a viable solution for applications facing intermittent connectivity, long propagation delays, and high error rates. These issues make it difficult to establish end-to-end paths among nodes and motivate the use of a store-and-carry forwarding model. A node keeps bundles (packets) in its buffer and forwards them opportunistically hop-by-hop until reaching the destination; therefore, the cooperation of the intermediate nodes significantly affects the network performance. However, to assume that all network nodes are cooperative could be impractical in some applications, and there could be some selfish nodes. Basically, a selfish node diverges from the policy of the routing protocol due to either resource scarcity or malicious behavior. This paper proposes DASH, a new reputation system to deal with selfish nodes. Rather than isolate the selfish nodes permanently from the network, we choose to merely avoid contact with them until they cooperate again. That way we give the selfish nodes an opportunity to change behavior, help in the bundle forwarding, and consequently improve the network performance. We simulated DASH in two protocols: Spray-and-wait and Maxprop. Results have demonstrated the effectiveness of DASH in detecting selfish behavior with low overhead, in increasing the overall network performance in terms of bundle delivery probability and latency.