【摘要】 针对认知无线电网络中多个次用户存在不同服务质量(quality of service, QoS)需求的频谱接入问题,提出了基于Dueling DQN (dueling deep Q-network)的分布式动态频谱接入方法。该方法通过与环境交互学习实现在次用户不掌握系统信道先验信息条件下动态获得最佳频谱接入策略,并以次用户碰撞次数以及成功接入信道次数分析比较所提出方法的性能。仿真结果表明,提出的方法在保护主用户不受干扰、满足多异质用户QoS需求的前提下,能够有效减少次用户间碰撞次数,提高次用户成功接入信道次数,相比随机接入与短视策略(myopic policy)频谱接入方法,该方法的碰撞次数分别降低60%和90%,其成功接入性能分别提高30%和50%。更多还原

【Abstract】 To solve the problem of dynamic spectrum access for multiple users with different quality of service(QoS) requirements in cognitive radio networks, this paper proposes a distributed dynamic spectrum access method based on Dueling DQN(dueling deep Q-network). The method can dynamically obtain the optimal spectrum access strategy through environment interactive learning when the secondary user does not have the prior information of the system channels. The performance of the proposed method is estimated by the number of collisions and successful accesses of secondary users. Simulations show that this method can effectively reduce the number of collisions between secondary users and improve the number of successful accesses of secondary users, under the premise of protecting the primary users from interference and meeting the QoS requirements of multiple heterogeneous users. Compared with random access and myopic policy methods, the collision times of this method can be reduced by 60% and 90%, respectively, and its successful access performance can be improved by 30% and 50%, respectively.更多还原