Performance Analysis of Reputation-based Mechanisms for Multi-hop Wireless Networks

Reputation-based mechanisms can be used to sustain cooperation among selfish users in a multihop wireless network. In these mechanisms, every node listens to its relaying neighbors, and the misbehaving users are punished by dropping a fraction of their packets, according to a tit-for-tat strategy. However, packet collisions prevent a node from recognizing a correct transmission, and this results in a distortion in the evaluated reputation. Thus, even if all the nodes cooperate correctly, a perceived defection may eventually lead to throughput loss due to retaliation. A possible way to mitigate this performance degradation is by adding a tolerance threshold to the pure tit-for-tat strategy, so that a limited number of defections will not trigger any punishment. In this paper, we propose a simple network model to study the impact of collisions on a reputation-based mechanism. Our results show that in a large ring network with uniform random traffic, a simple reputation-based scheme with an optimal choice of tolerance can achieve cooperation for any sustainable load, if the value for a packet to a node is sufficiently high.