北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2016, Vol. 39 ›› Issue (3): 64-69.doi: 10.13190/j.jbupt.2016.03.011

• 论文 • 上一篇    下一篇

无线Mesh网络功率控制与信道分配联合优化

石文孝, 王恩东, 王继红, 欧阳敏   

  1. 吉林大学 通信工程学院, 长春 130012
  • 收稿日期:2016-01-08 出版日期:2016-06-28 发布日期:2016-06-28
  • 作者简介:石文孝(1960-),男,教授,博士生导师,E-mail:swx@jlu.edu.cn.
  • 基金资助:

    国家自然科学基金项目(61373124)

Joint Power Control and Channel Assignment in Wireless Mesh Network

SHI Wen-xiao, WANG En-dong, WANG Ji-hong, OUYANG Min   

  1. College of Communication Engineering, Jilin University, Changchun 130012, China
  • Received:2016-01-08 Online:2016-06-28 Published:2016-06-28

摘要:

针对无线Mesh网络网关节点和网络链路承载的负载不均问题,择优选择网关节点,并设计链路权重,构建以网络加权吞吐量为优化目标的资源分配模型.在构建的资源分配模型下,提出一种基于Q学习和差分进化的联合功率控制与信道分配算法(QDJPCA).该算法通过获取功率控制的反馈结果,采用基于多重变异和自适应交叉因子的差分进化算法进行信道分配;针对每次迭代产生的信道分配结果,采用基于状态聚类和状态修正的Q学习算法实现功率控制.NS-3仿真结果表明,QDJPCA能够有效求解所提资源分配模型,在优先保证网关负载均衡和高负载链路吞吐量性能的基础上提升网络整体性能.

关键词: 无线Mesh网络, 功率控制, 信道分配, Q学习, 差分进化

Abstract:

To solve the problem that the load which is carried by gateway node and that the network links in wireless Mesh network is imbalanced, a resource allocation model was constructed to optimize network weighted throughput by selecting the best gateway node and designing the weight on each link. A joint power control and channel assignment algorithm (QDJPCA) was proposed based on Q learning and differential evolution under the resource allocation model. In this algorithm, the channel assignment is achieved by obtaining the feedback results of power control, using the multiple mutations and adaptive crossover rate based differential evolution algorithm. In each iteration, to achieve power control, the Q learning algorithm based on state clustering and state correction is utilized on the channel assignment result. NS-3 simulation shows that the QDJPCA can not only effectively solve the proposed resource allocation model, but also improve the overall network performance through preferentially ensuring the gateway load balancing and throughput performance of heavy load link.

Key words: wireless Mesh network, power control, channel assignment, Q learning, differential evolution

中图分类号: