Relay selection schemes relying on adaptive modulation and imperfect channel knowledge for cooperative networks

Cooperative MAC protocols have been shown to increase the attainable network throughput and to reduce the outage probability when perfect Channel State Information (CSI) is available either at the source station or at candidate relays. This paper investigates the realistically achievable network performance of relay selection schemes under imperfect CSI knowledge. Novel relay selection algorithms, based on a Markov Decision Process (MDP) framework, are proposed for mitigating the performance degradation introduced by imperfect channel knowledge. We demonstrate that the performance of cooperative MAC schemes crucially depends on the accuracy of the CSI knowledge. Furthermore, it is shown that the Markov framework allows the source station to predict the channel quality and hence to reduce the number of transmission errors, ultimately resulting in an improved network performance.