摘要
在5G D2D通信模式下,通过综合考虑中继节点在网络中所处的地理位置及中继节点处的信道增益,在多中继双向中继网络中提出一种基于CART决策树机器学习方法的优化中继选择策略RSBC。首先提出以地理位置和信道增益作为分类特征的两种分类方法,两种分类方法对应的系统性能增益不尽相同。为使最终选出的中继集能给系统带来最大性能增益,使用基尼系数和信息增益对两种分类方法进行了比较,并最终得到最优分类方式。在仿真中对系统的频谱效率和中断概率两个重要指标进行了分析比较,仿真结果证明了RSBC策略能够有效提升系统性能。
An optimal relay selection strategy named RSBC in multi-relay two-way relay network based on CART decision tree according to both considering geo-location and channel gain of relay nodes in 5 G D2 D network is proposed. Two classification methods based on geo-location and channel gain which will lead to different performance of the system are separately considered. In order to make the final selected relay bring the maximum performance gain to the system,two classification methods are compared using Gini coefficient and information gain,and finally the optimal classification method is obtained. In the simulation,frequency efficiency and outage probability,as two important parameters of the system,are analyzed and compared. The simulation results prove that the RSBC strategy can effectively improve the system performance.
作者
刘通
LIU Tong(Chongqing Vocational Institute of Engineering,Chongqing 402260,China;Chongqing University of Post and Telecommunications,Chongqing 400065,China)
出处
《中国电子科学研究院学报》
北大核心
2019年第10期1016-1021,共6页
Journal of China Academy of Electronics and Information Technology
基金
国家自然科学基金项目“基于多域环境感知的5G网络自组织机理与自优化方法研究”(61571073)
重庆市教委科学技术研究重大项目“基于数据驱动和人工智能的5G网络智能控制理论和方法”(KJZD-M201800601)
重庆工程职业技术学院科研课题“面向物联网智能应用的雾计算关键技术研究”(KJA201904)
