• cognitive radio;
  • fair scheduling;
  • collision statistics;
  • channel state prediction


In cognitive radio networks (CRNs), considering the randomness of primary users' (PUs) arrival and the nonideality of spectrum sensing performed by secondary users (SUs), the collisions between PUs and SUs are unavoidable. Frequent occurrences of collisions will strongly degrade PUs' and SUs' QoS. Therefore, collision statistics, such as the average number of collisions, is a very important performance metric in CRNs. If collision statistics are not considered in the scheduling scheme of the CRNs, some SUs will meet more collisions than the others, which cause unfair experiences among the SUs. Therefore, a fair scheduling scheme based on collision statistics is proposed in this paper, which can improve fairness across all SUs. First, the number of collisions is defined as an important fairness metric for each SU, and the scheduler dynamically adjusts the priorities of the SUs by periodically counting each SU's collision number. Then, a prediction algorithm, based on the continuous-time Markov chain model, is proposed to predict the idle probabilities of the available channels in the next slot. Considering both the priorities of the SUs and the idle probabilities of the available channels, a rational scheduling scheme will be achieved finally. Assuming the ordered hunt scheduling scheme applied by the primary system, the simulation results show that the proposed scheme can significantly improve the fairness across all SUs with little impact on spectrum utilization. Copyright © 2012 John Wiley & Sons, Ltd.