中文摘要：

频谱接入技术的关键是解决认知用户如何选择合适的空闲信道以及认知用户间如何实现频谱共享。在公共控制信道较难获得的情况下,基于部分可观测Markov决策过程（POMDP）的频谱预测算法,可以显著地提高系统的吞吐量;由于认知用户之间缺少接入信息的交换,使得多个认知用户在同一时隙接入同一信道的冲突概率较大。针对多用户竞争冲突问题,通过引入竞争反馈的偏差因子,并利用均衡的混合信道选择策略对认知无线网络动态频谱接入过程进行研究。通过大量仿真对认知用户的吞吐量和频谱利用率以及碰撞率进行分析,研究表明,均衡的混合信道选择策略可以有效地提高系统的吞吐量及频谱利用率。

英文摘要：

The key technology of the dynamic spectrum access is to solve the problem of choosing the proper channel and to realize the sharing of spectrum among the secondary users. When dedicated control channel was difficult to obtain, the spectrum prediction algorithm that based on the partially observable Markov decision process（POMDP） could significantly improve the throughput of the system. Lacking of exchange of control information among the secondary users, it increased the conflict probability of multiple cognitive users＇ accessing into the same channel at the same time slot. In order to solve the problem of competition conflicts among multiple users, this paper studied the process of cognitive wireless network dynamic spectrum access by introducing the deviation factor of competitive feedback and using a balanced hybrid channel selection strategy. In addition, it analyzed the following indicators： the user throughput, the spectrum utilization and the collision rate of different algorithms through simulation. Results show that a balanced hybrid channel selection strategy can effectively improve the user throughput and the rate of spectrum utilization.