基于双向强化学习与动态码率调节的无线mesh网络协议  

Bidirectional reinforcement learning and dynamic data rate adjustment based wireless mesh network protocol

作　　者：陈永建[1] 周艳[1] 崔海宁[2] 

机构地区：[1]肇庆学院电子信息与机电工程学院,广东肇庆526061 [2]吉林大学物理学院,长春130012

出　　处：《计算机应用研究》2017年第7期2118-2122,2139,共6页Application Research of Computers

基　　金：国家自然科学基金资助项目(61179055);广东省科技计划资助项目(2012B040303007);肇庆科技创新资助项目(2013G23)

摘　　要：针对现有无线mesh网络协议的用户体验质量(QoE)较差的问题,提出一种基于双向强化学习与动态码率调节的无线mesh网络协议。设计了兼容不同服务类型的无线mesh网络QoE度量框架和基于双向强化学习的无线mesh网络路由协议,并结合QoE感知的差异化报文调度策略与数据流源节点码率动态调节算法进一步优化终端用户的QoE质量。基于NS-2仿真平台的对比实验结果显示,协议可明显地提高无线mesh网络的QoE指标,同时具有较低的控制开销。Concerned the problem that the existing wireless mesh network（WMN） protocols show pool performance in quality of experience（QoE）, this paper proposed a bidirectional reinforcement learning and dynamic data rate adjustment based wireless WMN protocol. Firstly, it designed different types of services compatible QoE measure framework of WMN. Then, it designed bidirectional reinforcement learning based WMN routing protocol. Lastly, it combined QoE aware differential packets scheduling strategy and dynamic adjustment for data rate in source nodes of data flow to optimize the QoE quality of terminal user further. NS-2 simulation platform based compared experimental results show that the proposed protocol improves the QoE quality of wireless mesh network obviously, at the same time realizes a lower control overhead.

关 键 词：无线MESH网络 移动互联网 强化学习 用户体验质量 路由协议 码率调节 

分 类 号：TN915.04[电子电信—通信与信息系统]