Abstract
A channel-aware reservation-based MAC protocol for cognitive radio networks (CRNs) is proposed in this work. The proposed scheme allows secondary users (SUs) to send out reservation requests at the beginning of each time slot in a distributed fashion according to the sensing outcomes and their local channel state information (CSI). A channel-aware splitting algorithm is used to resolve the collision between the reservation requests of different SUs in a way such that the transmission time slot is always assigned to the SU with the best link quality. By exploiting multiuser diversity with the proposed scheme, we show that significant improvements in throughput can be observed. Furthermore, by modeling the presence and absence of the primary users (PUs) using a continuous time Markov chain, the spectrum sensing error and the reliability of the sensing outcomes across time are considered when designing the MAC policy. The effect of the data transmission and the spectrum sensing periods on the system throughput is studied via numerical simulations.