Dynamic rate adaptation in cognitive radio considering time-dependent channel access models

Recent studies showed that, in practical situations, the primary user (PU) access to the channel is better modeled as a time dependent random process. Taking this into account, we address the optimal rate adaptation problem of a cognitive radio (CR) link. A secondary user (SU) link detects an idle channel and starts the transmission with the goal of transmitting a given amount of data packets within a given time. During the transmission, and taking into account frame retransmission, the transmitter dynamically adapts the frames rate, from a finite number of available rates. If the PU accesses the channel, the SU immediately stops the transmission. The problem is formulated as an episodic Markov decision process (MDP). We show that selecting the best stationary policy (using the same rate for the whole transmission) can perform close to optimal.