超密集无线网络三维空间干扰分析及自组织干扰管理

Ultra-dense network deployment is a key technology which can accommodate massive connections and massive flow in the next generation mobile communication systems. However, wireless interference severely limits the performance of ultra-dense wireless networks (UDNs). Different from the traditional wireless networks, it is a challenge to depict and handle the interference in UDNs. In order to accurately analyze the interference status of UDNs, this project first measures the propagation parameters in the typical application scenarios of UDNs, and then constructs the wireless channel statistical models, which can depict the propagation characteristics of wireless signals in complex space. Based on the channel models, this project takes into full consideration the environment characteristics of UNDs to analyze the interference angle power spectrum and the spreading parameters, so as to comprehensively and accurately quantify the three-dimensional space interference status in different network regions. Guided by the interference analysis, this project adopts the reinforcement learning technique to investigate the local environment detection and self-optimization mechanism, and thereby design completely distributed and self-organizing interference management methods. By this way, the network interference can be efficiently coordinated, and the overall performance of the UNDs is enhanced. This project can provide theoretical foundation and technical support for the performance analysis and optimization of UDNs.

超密集网络布设是下一代移动通信系统应对巨连接和巨流量的关键技术，而无线干扰严重制约了超密集无线网络的性能。区别于传统的无线网络，超密集无线网络存在干扰难刻画、方法难实施的新挑战。为了准确分析超密集无线网络的干扰状态，本项目首先测量该网络典型应用场景中的传播参数，建立无线信道统计模型，刻画无线信号在复杂空间中的传播特性。基于无线信道模型，本项目在充分考虑环境特征的基础上分析超密集无线网络的干扰角度功率谱及其扩展参数，从三维空间角度全面、准确量化网络中不同区域的干扰状态。以干扰分析结果为指导，本项目进一步利用强化学习技术研究局部环境感知和自主优化机制，设计可完全分布式运行的自组织干扰管理方法，通过智能干扰管控有效协调网络干扰、提升网络总体性能。本项目的研究可以为超密集无线网络的性能分析和优化设计提供理论基础和技术支持。

本项目针对制约5G超密集无线网络性能持续提升的干扰问题展开系统深入的研究，围绕5G新频谱3.5GHz和28GHz开展大量外场实验，构建出城区高精度路径损耗模型，提出基于自组织神经网络的无线信道多径分簇算法，准确刻画出5G新频谱无线信号在密集网络复杂空间中的大尺度和小尺度传播特性。基于所提信道模型，本项目充分考虑站址、天线、障碍物等环境因素分析出超密集无线网络的干扰信号功率谱、到达角及其概率分布，从三维空间角度全面、准确量化网络中不同区域的干扰状态。以干扰分析结果为指导，本项目进一步针对各类典型超密集无线网络应用场景提出联合资源管理、网络部署、用户调度的自组织干扰管控技术，有效消除或协调了超密集无线网络中的同层和跨层干扰，显著提升了网络的谱效和能效，实现了网络容量的持续提升，为5G网络演进提供了关键技术支撑。在项目执行期内，共发表SCI期刊论文12篇（均为中科院二区及以上）、国际会议论文9篇，申请发明专利6项（其中授权4项），指导学生获得2021年研究生电子设计竞赛国家一等奖和最佳商业价值奖，以及2020年第六届中国国际“互联网+”大学生创新创业大赛陕西赛区金奖等奖项。