多天线频谱感知与接入模型协议研究

发布时间：2018-08-27 06:32

【摘要】：无线技术由于可以灵活方便地接入通信网络而备受青睐。据Erikson全球移动公司报道,2013年手机用户数达到了75亿,这一数字已超过当前70亿的世界人口数。无线通信已成为服务技术的主流且规模正在不断扩大,人们不得不在公用网内开发云端技术来处理大数据量的传输。随着网络用户向移动互联网的转移和物联网技术的发展,无线传输的容量和可靠性被赋予了更高的要求。然而如何提高频谱利用率、采用有限的频谱资源承载这些业务是无线通信技术进一步发展必须要解决的首要问题。固定的频谱分配政策是造成频谱利用率低下的主要原因。2003年美国联邦通信委员会(FCC)和国际电信联盟(ITU)等频谱机构发现各地区的频谱利用率非常低下。为了提高认知系统的频谱使用效率和对频谱环境的适应能力,无线网络需要具有传感能力。1999年美国提出了通过谱传感利用频谱的认知无线电标准,采取动态频谱管理技术允许非授权用户伺机接入授权用户的空闲频谱而实现对频谱的有效利用。据称该技术可以完全解决频谱匮乏的问题,已被认为是未来无线网络的核心技术之一。动态利用频谱需要解决频谱感知和频谱接入两个方面的问题。前者要求系统对空闲频谱实施快速而准确的检测；后者要求用户能自适应地接入频谱空洞,并在发现主用户的情况下快速地切换到其他信道。隐终端和无线衰落等不利的环境因素使得用户在相互覆盖范围内无法发现潜在的通信冲突,使得感知性能大打折扣；同时当网络负载较大时,频谱共享系统的服务质量会下降。本文对这两个方面涉及的理论与技术问题进行深入探索,主要工作如下：首先,介绍认知无线电的研究背景,说明当前频谱危机和授权频谱利用率低下的现实。然后对认知无线电技术的概念、特点、各种标准、项目和应用前景进行了阐述。第二,研究了多天线分集合并技术和频谱感知技术,并提出一种多天线的智能谱感知方法,该方法融合了匹配滤波器检测、能量检测和循环平稳检测算法。通过假设信道存在高斯噪声和平坦Nakagami衰落而建立了相关的二元假设检验模型,并推导出瞬时检测概率的封闭表达式；根据SC、EGC、MRC三种多天线合并方式下的幅度增益的随机统计特性,分别解析了瞬时信噪比对应的概率分布。通过将瞬时检测概率在其概率密度函数上进行平均而导出精确的或近似的平均检测概率表达式,同时根据各检测算法的事件概率计算出平均检测时间的表达式。其后利用MATLAB的数值计算工具分析了检测概率和检测时间的ROC特性。第三,研究动态频谱接入的协议和模型。提出了一种基于服务期损失制轮询调度的有中心混合频谱接入方案,次用户根据感知结果和干扰温度判断主用户的活动状态,并概率式地接入信道。通过构建Nakagami信道下的MAC层排队模型,解析了次用户的平均服务率的封闭表达式。为了保证主用户的服务质量,规定主用户具有最大容忍延时,以此为约束条件分析了平均服务率的理论最大值。由于次用户的业务类型往往多样化,该部分又进一步研究了次用户系统具有多队列排队结构的情形,考虑了两种调度方案：第一种为概率式调度方案,采用拉格朗日乘子法得到了各个队列的接入概率；第二种是损失制服务轮询调度机制,在传统的门限服务、限定1服务和完全／门限两级优先级服务三种的轮询模型中假设有顾客离开,然后采用嵌入式Markov链和概率母函数的分析方法建立了与此调度方案相吻合的数学模型及其函数关系式,并精确解析出系统特性参数(平均排队队长、平均循环周期、平均等待时延)的表达式。其后利用MATLAB的数值计算工具分析了以上参数的ROC特性。第四,基于OPNET软件,根据前节的多天线感知和轮询频谱接入方案设计了认知无线电感知接入仿真平台,包含三个方面：在物理层,用Simulink搭建信号衰落、加性高斯信道、产生PU信号类型等过程；在MAC层,用OPNET搭建节点模块、数据源产生模块、队列模块；物理层和MAC层通过OPNET和MATLAB的联合仿真引擎接口进行信息交互。然后利用该实验平台对设定的仿真场景下的频谱利用率、平均排队队长和平均等待时延等相关参数进行了测试研究。
[Abstract]:Wireless technology is favored for its flexible and easy access to communication networks. According to Erikson Global Mobile, the number of mobile phone users reached 7.5 billion in 2013, which exceeds the current 7 billion world population. With the transfer of network users to the mobile Internet and the development of the Internet of Things technology, the capacity and reliability of wireless transmission have been given higher requirements. The first problem to be solved.
Fixed spectrum allocation policy is the main reason for low spectrum utilization. In 2003, the Federal Communications Commission (FCC) and the International Telecommunication Union (ITU) found that the spectrum utilization rate in various regions is very low. In 1999, the United States proposed a spectrum sensing cognitive radio standard, which allows unauthorized users to access the free spectrum of authorized users by dynamic spectrum management technology. It is said that this technology can completely solve the problem of spectrum scarcity and is considered to be the future wireless network. One of the core technologies of network.
The former requires the system to detect the idle spectrum quickly and accurately, while the latter requires the user to access the spectrum holes adaptively and switch to other channels quickly when the primary user is found. Environmental factors make it impossible for users to discover potential communication conflicts within the scope of mutual coverage, which greatly reduces the perception performance. At the same time, when the network load is large, the quality of service of spectrum sharing system will decline.
Firstly, the research background of cognitive radio is introduced, and the current spectrum crisis and the low utilization of authorized spectrum are illustrated.
Secondly, the multi-antenna diversity combining technique and spectrum sensing technique are studied, and a multi-antenna intelligent spectrum sensing method is proposed, which combines matched filter detection, energy detection and cyclostationary detection algorithm. According to the random statistical characteristics of amplitude gain of SC, EGC and MRC, the corresponding probability distribution of the instantaneous signal-to-noise ratio is analyzed. The accurate or approximate average detection probability is derived by averaging the instantaneous detection probability on its probability density function. At the same time, the expression of average detection time is calculated according to the event probability of each detection algorithm. Then the ROC characteristics of detection probability and detection time are analyzed by using MATLAB numerical calculation tool.
Thirdly, the protocol and model of dynamic spectrum access are studied. A centralized hybrid spectrum access scheme based on polling scheduling with loss of service is proposed. Secondary users judge the active state of primary users according to sensing results and interference temperature, and access the channel probabilistically. A closed expression of the average service rate of secondary users is given. In order to guarantee the quality of service of primary users, the maximum tolerant delay of primary users is specified, and the theoretical maximum of the average service rate is analyzed with this constraint. In the case of structure, two scheduling schemes are considered: the first is probabilistic scheduling scheme, and the access probabilities of each queue are obtained by using Lagrange multiplier method; the second is loss-based service polling scheduling mechanism, which is assumed in three polling models: traditional threshold service, limited one service and full/threshold two-level priority service. Then, the mathematical model and the function relation are established by using the method of embedded Markov chain and probability generating function, and the expression of the system characteristic parameters (average queue length, average cycle period, average waiting delay) is obtained accurately. The ROC characteristics of the above parameters are analyzed.
Fourthly, based on OPNET software, a cognitive radio sensing access simulation platform is designed according to the multi-antenna sensing and polling spectrum access scheme in the preceding section, which includes three aspects: in the physical layer, Simulink is used to build signal fading, additive Gaussian channel, and PU signal type; in the MAC layer, OPNET is used to build node module and data source generation. Module, queue module, physical layer and MAC layer interact with each other through the joint simulation engine interface of OPNET and MATLAB. Then the spectrum utilization, average queue length and average waiting delay in the simulation scenario are tested and studied.
【学位授予单位】：云南大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN925

【参考文献】