基于DAQL算法的动态频谱接入方案  

Dynamic spectrum access based on DAQL

作　　者：吴启晖[1] 刘琼俐[1] 

机构地区：[1]解放军理工大学通信工程学院,江苏南京210007

出　　处：《解放军理工大学学报（自然科学版）》2008年第6期607-611,共5页Journal of PLA University of Science and Technology(Natural Science Edition)

基　　金：国家863计划资助项目(2007AA01Z267);国家973计划资助项目(2009CB3020402)

摘　　要：针对传统的动态频谱接入方案一般没有考虑自主性,不具备普适性这一缺点,提出了一种基于双动作Q学习算法DAQL(double action Q-learning)的频谱接入方案,该方案将DAQL引入到多授权用户存在的环境下频谱接入问题中,用以降低接入未知频谱环境时的冲突概率。仿真结果表明,提出的方案与随机接入方案相比,不但有更小的冲突概率,而且能动态适应环境的变化,适合认知无线电的需要。As for the traditional dynamic spectrum access schemes that are not fit to all kinds of instances and do not consider the ability of self-determination, a scheme based on Double action Q-learning （DAQL） was proposed to solve the problem of dynamic spectrum access. This scheme could reduce the collision probability when accessing unknown spectrum by introducing DAQL into the scenario where there were many licensed users. Simulation results show that this scheme has lower collision probability than random access scheme, and can be adaptive to the change of environment so that it can satisfy the need of cognitive radio.

关 键 词：强化学习 Q学习 双动作Q学习算法 冲突概率 

分 类 号：TN929.5[电子电信—通信与信息系统]