摘要
针对认知无线电网络中认知用户(CR)的机会频谱感知及接入问题,提出了一种基于分布式部分可观测马尔科夫决策过程(Dec-POMDP)的多用户频谱接入算法。在该模型框架下,相邻CR用户通过交换接入策略,以区域策略梯度方向为基准,对各个CR用户的接入策略做出调整,从而得到最优联合接入策略。仿真结果表明:该算法有效降低了授权用户的容量损失,提高了空闲频谱的利用效率,能够更有效地做出接入决策。
In the Cognitive Radio(CR) network,the opportunistic spectrum sensing and access is of paramount importance to the primary user's capacity.This paper propose a multi-user Decentralized Partially Observable Markov Decision Process(Dec-POMDP) CR spectrum access algorithm.In this framework,adjacent CR users exchange access policies with neighbors,and according to policy gradient,each local CR user adjusts its own access policy till the system of CR network obtains an optimum joint access policy.Simulation results show that this algorithm can lower the capacity loss of authorized user efficiently and improve spare spectrum efficiency,moreover,it can make access decision more efficiently.
出处
《信息与电子工程》
2010年第6期720-725,共6页
information and electronic engineering
基金
国家自然科学基金资助项目(60972024)
教育部博士点基金资助项目(20090071120087)
专用集成电路与系统国家重点实验室开放课题(20080402)
关键词
认知无线电
可观测马尔科夫决策过程
策略梯度
频谱分配
Cognitive Radio
Partially Observable Markov Decision Process
policy gradient
spectrum allocation