【作者】 滕志军；

【导师】 赵春晖；

【作者基本信息】 哈尔滨工程大学 ， 信号与信息处理， 2012， 博士

【摘要】 无线资源管理的目标是:在满足用户QoS要求的条件下，在有限的带宽上最大限度地提高频谱效率和系统容量，同时避免网络拥塞的发生。认知无线电技术通过从时间和空间上充分利用空闲频谱有效提高频谱利用率。准确的感知是认知无线电技术使用的前提，为了避免对授权系统造成有害干扰，认知系统首先要感知频段，发现频谱机会。认知无线电的学习能力是使它从概念走向实际应用的真正原因，有了足够的人工智能，它就可能通过吸取过去的经验来对实际的情况进行处理，过去的经验包括对频段上的使用情况，干扰情况，信号特征，解决方法等方面的内容。认知无线电技术赋予无线电设备根据频带可用性、位置和过去的经验来自主确定采用哪个频带的功能。所以研究认知无线电中频谱管理策略及基于该技术的MIMO系统容量分析更具有现实的意义。本文旨在研究认知无线电频谱检测及分配算法，主要工作包括:首先，研究了基于不同间隔的频谱检测算法。在等间隔频谱检测算法和现有文献基础上，衍生出非线性可变周期频谱检测算法，在对主用户的检测中，采用可变周期与采用固定周期相比较，减少了检测损失时间，有效降低了检测成本，同时实现了对频谱检测成本的有效控制。引入二元场论(阴阳理论)，将主用户与认知用户假定为阴阳二体，借助太极哲学的思想——阴阳之间的比例是动态变化的，通过相互调整以达到平衡，阴阳之间的动态优化平衡。通过寻求阴阳之间的平衡点，就可以在两种对立的状态之间做出最优决策。仿真结果表明，基于二元场论的频谱检测算法可以更加有效地降低检测成本。其次，研究基于博弈论的认知无线电频谱分配算法。建立基于博弈论的认知无线电频谱分配模型，提出基于潜在博弈的认知无线电频谱分配算法，该算法能够在较短时间内收敛到纳什均衡状态，潜在函数值也随着算法的收敛逐步提高并达到最大，系统中各用户及整个系统的SIR水平得到明显改善，实现了提高频谱利用率的目的。综合考虑了认知无线电系统频谱分配问题的动态性和复杂性，结合博弈论提出了一种适合于认知无线电系统的基于多次博弈的动态频谱分配算法，通过数学分析，求出纳什均衡点，并证明了该算法中纳什均衡点的存在性和唯一性。同时提出了一种适合于认知无线电系统的基于多次博弈的功率控制模型，仿真结果表明，基于该模型的算法收敛性较好，经过7次左右的迭代即可收敛，既保证了不同用户对输出SIR的要求，又有效地降低了用户发射功率的消耗，在尽可能降低功率的前提下给用户的效用带来更高的保证，实现了资源的最优化配置。最后，研究了认知MIMO系统中的信道容量和两种相关模型下的相关系数与容量的关系。通过研究分布式MIMO系统的信道容量，建立了能同时反映小尺度衰落和大尺度衰落的分布式MIMO系统信道模型，给出了分布式MIMO系统的非相关和相关容量的计算公式。引用MIMO遍历信道容量经典公式，建立空间相关MIMO信道的矩阵模型，在此基础上推导出空间相关MIMO信道遍历容量的上界，得出常量相关模型的信道容量上界以及容量和相关系数的关系，在指数相关模型中得出容量和概率密度的关系，为了摆脱各种因素的制约，针对如何降低分析和计算的复杂度进行研究，从而使常量相关、指数相关模型被更加广泛使用。更多还原

【Abstract】 Radio resource management goal is to improve spectrum efficiency and system capacityto the hilt with a limited bandwidth while avoiding network congestion under the condition ofmeeting the demand of user QoS. Cognitive Radio technology improves spectrum utilizationeffectively by making full use of idle spectrum in time and space. Spectrum sensingtechnology has become the basis and premise of Cognitive Radio technology, for cognitivesystem should perceive frequency band and find the spectrum opportunity to avert harmfulinterference to licensed systems. Having enough artificial intelligence, Cognitive Radio candeal with practical matters based on lessons, which include service condition, disturbedcondition, signal features as well as solutions of frequency band learned in the past. Namely,learning ability makes it into practical application from concept. Cognitive Radio technologyenables the radio equipment, by using the band usability, position and previous experience tochoose the appropriate frequency band independently. So the study of cognitive radiospectrum management strategy and the analysis of MIMO system capacity based on thistechnology are much more realistically significant.This paper aims to study cognitive radio spectrum detection and distribution algorithm.The major contents in this paper include:Firstly, spectrum detection algorithm based on the different interval is studied. Anonlinear variable period spectrum detection algorithm is derived from the equal intervalspectrum detection algorithm and the existing literature. The variable cycle and fixed cycledetection algorithm are compared in the primary user’s detection, which reduces the detectingtime and the test cost. The method controls the spectrum test cost effectively. The main usersand cognitive users are assumed to be Yin and Yang two bodies, introducing the binary fieldtheory（yin-yang theory). The thought of Tai chi philosophy--the ratio between the Yin andYang changes quite dynamically. The dynamic optimization balance can be achieved throughthe mutual adjustment, is cited. The optimal decision between two opposite states can bemade by seeking the balance point of Yin and Yang. The simulation results show thatspectrum detection algorithm based on yin-yang theory can be more effective to reduce testcost. Then, Cognitive Radio spectrum allocation algorithm based on game theory is studied.The cognitive radio spectrum allocation model is established and the cognitive radio spectrumallocation algorithm based on the potential game is proposed. The algorithm can converge tothe Nash equilibrium state in a relatively short period and potential function value graduallyrises till reaches maximum with the algorithm convergence. Thus, the SIR levels of users aswell as the whole system have been improved obviously and the purpose of improving thespectrum utilization is realized. A dynamic spectrum allocation algorithm based on multiplegame is proposed after considering the dynamics and complexity of Cognitive Radio systemspectrum allocation problem comprehensively and combining with game theory. A dynamicspectrum allocation algorithm based on multiple game is proposed after considering thedynamics and complexity of Cognitive Radio system spectrum allocation problemcomprehensively and combining with game theory. The Nash equilibrium point is worked outthrough mathematical analysis and its existence and uniqueness are proved. A power controlmodel based on multiple game is also proposed, which is suitable for Cognitive Radio system.The simulation results show that the algorithm based on the model has good convergencewhich can be reached after about seven times’ iteration, thus ensuring the different users’output SIR request, lowering the user emission power consumption effectively and efficientlyguaranting the user’s utility under the premise of reducing the power as far as possible. That’sto say, the optimization of resource allocation is realized.Finally, the channel capacity of multiple-input multiple-output (MIMO) system and therelationship between correlation coefficient and the capacity of two related model are studied.Through studying the channel capacity of the distributed multiple-input multiple-output(MIMO) system, the distributed system channel model which can reflect both the small scalefading and large scale decline is established and the computational formula of unrelated andrelated capacity is presented. Introducing classic formulas of MIMO ergodic channel capacity,the matrix model of spatial correlation MIMO channel is set up to derivative the upper boundof the ergodic channel capacity. The channel capacity upper bound and the relationshipbetween the capacity and the correlation coefficient of constant related model, as well as therelationship between the capacity and probability density of index related model are alsoderived. In order to get rid of the restriction of various factors, researches are conducted to reduce the complexity of analysis and calculation so as to make constant related model andindex related model to be used more widely.更多还原