The Study Of QoS Provisioning Mechanism In Cognitive Radio Networks

Researches show that Cognitive Radio(CR) technology plays an important role in improving spectrum efficiency. However, it also makes the primary users(PUs) and secondary users(SUs) quality of service(QoS) being not guaranteed. Hence, we still will face the great challenges in strictly ensuring traffics' QoS in Cognitive Radio Networks(CRNs). So far, although the traditional QoS provision mechanisms(admission control based on channel reservation and static probability) can guarantee the users' Qo S requirement, the control precision is insufficient, and all are based on the assumption of perfect channel sensing, and even worse, it remains to be improved in terms of improving the spectrum utilization. Because admission control strategy is based on observed state space, how to map the system's real state space into observation space becomes the main challenge in reality with imperfect sensing(i.e., partially sensing and wrongly sensing).In view of these challenges and shortages, in this paper, we provide an admission control strategy with higher control precision and better system performance—Jointly Admission and Eviction Control with Dynamic probability(D-JAEC)—under the condition of imperfect sensing. The target of this strategy is to strictly guarantee the Qo S requirement with respect to secondary users' dropping probability, average handoff delay and primary users' collision probability while maximizing the throughput of a cognitive system. We first develop a three-dimension continuous-time Markov chain(CTMC) to analyze the system performance metrics, and then computing the optimal admission probability and elimination probability by solving a non-linear programming problem of Partially Observation Markov Decision Process(POMDP). Compared with the traditional QoS provision mechanisms, our scheme outperforms a lot.