面向轨道交通宽带无线通信的大规模多智能体认知无线频谱管理的关键性技术研究

项目名称： 面向轨道交通宽带无线通信的大规模多智能体认知无线频谱管理的关键性技术研究

项目编号： No.61471252

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 无线电电子学、电信技术

项目作者： 吴澄

作者单位： 苏州大学

项目金额： 80万元

中文摘要： 对于轨道交通公众宽带通信，采用认知无线电技术提升其频谱资源利用率极具实用价值。可是，认知无线网络中包含大量用户,和认知无线用户之间存在合作和竞争行为是轨道交通认知无线网络的两个重要基本特征。这也是导致整体网络动态性能低下的关键性原因。本项目立足于轨道交通认知无线网络呈大规模化的特殊性，分析认知无线电个体用户数量和行为与整体网络性能之间动态关联性，解析大规模认知无线用户智能体在轨道交通通信中协同工作的运作机理，运用已取得的多智能体价值函数近似理论的研究成果，建立适合大规模认知无线用户交互的近似状态空间构架，从而改善目前轨道交通认知无线网络中人工智能技术处理大规模化的能力缺失且效率低下的现状。本研究的成果有望进一步揭示大规模应用性认知无线网络的学习进化演变规律，对目前迫切需要解决的应用认知无线网络处理大规模复杂通信问题有着重要的参考价值，同时对完善认知无线电智能理论有一定的理论意义。

中文关键词： 大规模认知无线网络；多智能体系统；频谱管理；函数近似；轨道交通宽带无线通信

英文摘要： In terms of public broadband communications in rail transportation, using cognitive radio technology to enhance the efficiency of spectrum usage are greatly valuable. However, there are two fundamental characteristics for cognitive wireless network in rail transportation: it contains a large number of cognitive radio users, and it has cooperation and competition behaviors between these users. These characteristics also result in the issue of bad performance for overall network dynamics.In particular, our research just start from the understanding on the characteristics of cognitive wireless network in rail transportation. We first analyze quantitative correlation between the amount and behaviors of individual users and dynamic network performance.We then use multi-agent reinforcement learning mechanism to resolve the cooperation of cognitive radio users in large-scale communication system in rail transportation. Finally, based on our past research about Value Function Approximation theories, we establish the approximation framework of multiple users' cooperation in large-scale cognitive wireless network.The research will improve the efficiency of applying artificial intelligence techniques to resolve spectrum management of cognitive wireless network in rail transportation. The results from our research is to reveal the evolution process of learning policy in large-scale cognitive wireless networks. It can also be used as a reference of applying value function approximation technique to solve large-scale complex communication problems. The research, as part of cognitve radio research, will be theoretical significant to improve in cognitive radio intelligence.

英文关键词： Large-scale Cognitive Wireless Network;Multi-agent System;Spectrum Management;Function Approximation;Broadband Wireless Communication in Rail Transportatoin