一种基于马尔可夫决策过程的认知无线电网络传输调度方案计划计划atr.pdf

第31 卷第8 期 电 子 与 信 息 学 报 Vol.31No.8 2009 年8 月 Journal of Electronics Information Technology Aug..2009 一种基于马尔可夫决策过程的认知无线电网络传输调度方案 朱 江① 徐斌阳② 李少谦① ①(电子科技大学通信抗干扰国家级重点实验室 成都 610054) ②(上海贝尔阿尔卡特创新中心 上海 201206) 摘 要：该文提出了一种适用于认知无线电网络的跨层传输调度方案，即满足掉包率约束的前提下最小化平均功率 消耗。此方案被建模为约束马尔可夫决策过程(MDP)。采用拉格朗日乘子法求解此MDP，并且提出了一种黄金分 割乘子有哪些信誉好的足球投注网站法。提出两种简化方法，即状态聚合以及行动集缩减来解决维灾问题。仿真结果显示简化方法对该方案 的性能影响很小，且该方案的平均功耗最低。 关键词：认知无线电；马尔可夫决策过程；跨层设计；传输调度 中图分类号：TN92 文献标识码：A 文章编号：1009-5896(2009)08-2019-05 A Transmission and Scheduling Scheme Based on Markov Decision Process in Cognitive Radio Networks ① ② ① Zhu Jiang Xu Bin-yang Li Shao-qian ① (National Key Laboratory of Communications, University of Electronic Science and Technology of China, Chengdu 610054, China) ② (Research and Innovation Center of Alcatel Shanghai Bell, Shanghai 201206, China) Abstract: A cross-layer transmission and scheduling scheme of average power minimization in cognitive radio networks under the constraint of packet drop probability is addressed. The scheme is formulated by constrained Markov Decision Process (MDP). Lagrangian multiplier approach is used to solve the MDP, and a golden section search method is proposed to find the multiplier. Two simplifying methods, namely, state aggregate and action set reduction are employed to cope with the curse of dimensionality. Simulation results show that simplifying methods have little influence on the performance of the scheme and average power consumption of the sche