基于强化学习的频谱决策与传输算法  

Spectrum Decision and Transmission Algorithm Based on Reinforcement Learning

作　　者：江虹[1] 伍春[1] 刘勇[1] 

机构地区：[1]西南科技大学信息学院,绵阳621010

出　　处：《系统仿真学报》2013年第3期565-570,共6页Journal of System Simulation

基　　金：国家自然科学基金(61072138);国防基础科研计划(B3120110005);国家973计划项目(2009CB320403)

摘　　要：在认知无线电(CR)通信中,各信道可能具有不同的带宽、干扰强度和主用户冲突概率,如何据自身业务特性选择最佳信道和传输策略是系统设计的关键问题之一。提出一种基于Q学习的在线学习算法,用于解决多用户多信道CR系统中的信道选择与自适应传输问题。在不知道信道状态信息和主用户业务特性情况下,通过在线学习,获得各种环境下的最佳频谱选择与自适应传输策略。为验证所提方法的有效性,采用随机频谱选择算法和最小干扰频谱选择算法与所提方法进行比较。仿真结果表明,提出的方法通过在线学习实现了认知无线电的自适应控制,能够有效增加认知无线电的通信性能。In cognitive radio （CR） communications, each channel may have different properties such as bandwidth, interference strength and PU conflict probabilities, etc. It＇s one of key problems how to select the best channel and transmission strategy according to their own service types. An online learning algorithm based on Q-Learning was proposed to solve the problem of channel selection and adaptive transmission for multi-user and multi-channel CR system. The scheme, although did not know the channel state and PU traffic characteristics, could achieve the best spectrum selection and adaptive transmission strategy through learning interactive experience with environment. In order to verify the effectiveness of the scheme, it was compared with a random spectrum selection and minimum interference selection algorithm. The simulation results show that the online learning scheme realizes the adaptive control of CR, and increases the communication performances of CR.

关 键 词：认知无线电 频谱决策 Q学习算法 自适应传输 

分 类 号：TN915[电子电信—通信与信息系统]