Skip to Main Content
In distributed wireless networks without centralized control, each user is its own authority to maximize its own performance. This distributed characteristic provides the users with the incentives of greedy competitions for the network resources such as bandwidth and transmission time. Such competitions deteriorate the system performance dramatically and result in low non-cooperative performances. Ensuring cooperation among selfish users can improve the performances and thus becomes an important issue for such wireless networks. In this paper, a Cartel Maintenance repeated game framework is proposed to enforce the cooperation among selfish users. The soul of Cartel Maintenance is to construct contracts among independent individuals for cooperative benefits and non-cooperative punishment, so as to limit inefficient competitions. In the proposed framework, a trigger-punishment game rule is designed to encourage the users to follow the cooperative strategy. First, all users agree to cooperate. Then in the following time slot, if users observe the others play cooperatively, cooperation will be played. If some users detect that others may defect based on the observed information, these users will play punishment phases for a certain period of time. This punishment is optimized so that the gain obtained by the deviation users is outweighed by the future punishments. Therefore, no user has the incentives to deviate, and the cooperation among selfish users is enforced. Then, the framework is employed to the multiple random access scenarios in wireless networks with selfish users, where the closed-form optimal solutions of cooperation enforcement are derived. The simulation results show that the proposed scheme can achieve significant performance gains over the non-cooperation scheme by having enough punishment threat to keep the cooperation among users.
Date of Publication: May 2008