Throughput Maximization in Cognitive Radio System with Transmission Probability Scheduling and Traffic Pattern Prediction

In this paper, we propose a novel transmission probability scheduling (TPS) scheme for the opportunistic spectrum access based cognitive radio system (OSA-based CRS), in which the secondary user (SU) optimally schedules its transmission probabilities in the idle period of the primary user (PU), to maximize the throughput of the SU over a single channel when the collision probability perceived by the PU is constrained under a required threshold. Particularly, we first study the maximum achievable throughput of the SU when the proposed TPS scheme is employed under the assumption that the distribution of the PU idle period is known and the spectrum sensing is perfect. When the spectrum sensing at the SU is imperfect, we thoroughly quantify the impact of sensing errors on the SU performance with the proposed TPS scheme. Furthermore, in the situation that the traffic pattern of the PU and its parameters are unknown and the spectrum sensing is imperfect, we propose a predictor based on hidden Markov model (HMM) for the proposed TPS scheme to predict the future PU state. Extensive simulations are conducted and show that the proposed TPS scheme with the HMM-based predictor can achieve a reasonably high SU throughput under the PU collision probability constraint even when the sensing errors are severe.