高铁通信环境下的网络选择及认知频谱分配联合算法研究

With the increasing demand of the data-carrying capaciy, the high-speed railway mobile communication faces the problem of limited type of service, un-guaranteed QoS and low transmission rate, due to the scarcity of licensed spectrum resource. This project will study a joint algorithm for the network selection and cognitive spectrum allocation in the high-speed railway communication. The different attribute factors of railway broadband communication scenarios will be characterized, explained and processed through AHP and GRA methods, so that the scenario characteristic values can be evaluated and measured rationality. Besides, the features of cognitive radio resource are modeled and investigated using Q-learning method, then a spectrum ranking and decisioning scheme which can gurantee the QoS of cognitive users will be established. Moreover, a new joint utility function which can reflect the high-speed railway’s operating scenario and QoS level will be built up by the utility theory, the best cooperative way and the cooperative condition between network selection and spectrum allocation will be explored, the mathematical relationship between the spectrum price per unit and the number of QoS levels, the numbers of CR users will be conducted. Furthermore, a repeated-game inspired spectrum allocation algorithm will be studied to guarantee the QoS provision and fairness. The goal of the project is to achieve a promising joint algorithm for railway network selection and spectrum allocation with QoS provision, fast convergence, low complexity, the end-to-end delay below 0.5s. The achievements of the project will provide theoretical foundations and technical support for efficient network construction and evolution to the high-speed railway broadband mobile communication network.

针对高速铁路移动通信的数据承载量剧增而现有系统因授权频谱匮乏导致业务类型受限、传输速率低的问题，本项目研究高铁通信环境下的网络选择及认知频谱分配联合算法。用层次分析法和灰度关联法表征、解释和处理各类场景属性因子信息，合理评价及度量高铁宽带移动通信网络场景特征量；采用Q-learning算法，对认知频谱资源属性特征展开建模研究，建立能够实现认知用户QoS保障的频谱排序及判决机制；拟研究效用理论，构建能反映高铁运营场景、QoS级别等因素的新的联合效用函数，掌握网络选择及频谱分配的最佳协作方式、协作发生条件，建立频谱单价与认知用户业务等级数、QoS级别数以及认知用户数之间的数学关系；研究具有QoS保障且兼顾公平性的重复博弈频谱分配算法。旨在提出收敛速度快、复杂度低、端到端传输时延不高于0.5s的网络选择及频谱分配联合算法，为高铁宽带移动通信网络的构建及演进提供理论基础和技术支撑。

为解决高速铁路移动通信的数据承载量剧增导致的授权频谱匮乏、业务类型受限、传输速率低的问题，本项目对高铁通信环境下的网络选择及认知频谱分配联合算法展开研究。研究内容如下：.（1）高速铁路宽带移动通信网络场景特征量的评价及度量。测试表明，高铁网络覆盖需要考虑基站间距、天线挂高、天线方位角和下倾角等基本工程参数。针对高铁移动用户在异构无线网络中的优化网络选择问题，用户QoS属性参数包括：当前网络可用带宽、时延、抖动、丢包率和价格，应根据用户业务类型设计不同的回报函数。.（2）CR网络选择及频谱分配联合机制。提出基于二分图的联合优化匹配算法，在网络端采用灰度关联分析构建异构无线网络的备选集合和判决矩阵；在用户端，将备选方案分为不同的等级并进行优先级排序。算法以联合优化空闲网络频谱资源利用率和用户在网络间切换率为目标，解决用户和备选网络资源之间的分配选择问题。算法通过用户端和网络端的共同决策，优化了吞吐量，均衡了网络负载。.（3）基于效用函数的分级频谱分配。采用二次函数表示次用户共享频谱过程中的效用，建立效用函数与不同主用户频谱之间的频谱替代性参数、主用户的频谱占用情况、次用户租借主用户的频谱数量和次用户的频谱传输效率等参数之间的函数关系；基于二次边际成本函数和博弈模型，实现了优先级保障的认知用户频谱分配。.（4）具有QoS保障且兼顾公平性的频谱博弈算法。提出了认知频谱共享动态博弈算法。引入次用户（列车用户）优先级机制，优先满足次服务中关键任务用户和高实时性业务用户的频谱需求；考虑主服务的收益和次服务的满意度，得到满足稀缺资源分配属性的频谱价格；利用非合作博弈建模次用户间的竞争过程，保证了次用户的公平性和效用最大化。.项目提出了具有QoS保障且兼顾公平性的频谱博弈算法，解决了不同业务优先级高铁用户的爆炸式业务需求与稀缺频谱资源的矛盾，为高铁宽带移动通信网络的演进提供理论基础和技术支撑。