To achieve full cooperative diversity gains while still obtaining spectral and energy efficiency, cooperative communications with relay selection schemes, i.e., only the best relay is selected from multiple relaying candidates to cooperate with the communication, have been extensively studied in recent research. In this paper, we review the recent research on optimal relay assignment for cooperative communications, and investigate the use of cooperative communications with adaptive relay selection for soft QoS provisioning in resource-constrained wireless sensor networks. We propose EEARS, an energy-efficient adaptive relay selection scheme, which is based on a multi-agent reinforcement learning framework. In EEARS, optimal relays, in terms of outage probability, spectral and energy efficiency, are selected distributedly from multiple relaying candidates to participate in the communication, without the needs of prior knowledge of the wireless network model and centralized control. Simulation results show that EEARS fits well in dynamic environments, and is effective in improving the satisfying level of soft QoS provisioning for WSNs, i.e., increasing the spectral and energy efficiency, and reducing the amount of time that QoS violation occurs, by exploiting spatial and time diversities. Furthermore, compared with schemes using fixed transmitting power, EEARS can achieve a higher energy efficiency by varying the transmission power level according to wireless channel conditions.
展开▼
