基于压缩感知的宽带频谱检测方法研究

发布时间：2018-06-14 07:43

本文选题：认知无线电 + 压缩感知　；参考：《北京邮电大学》2014年博士论文

【摘要】：全球无线通信技术的发展日新月异,最有代表性的是蜂窝移动通信系统、无线局域网、卫星通信、短距离无线通信及集群通信等。在目前固定的频谱分配方式下,使得能用于再分配和使用的频率变成了极为稀缺的资源,同时,这些已向授权用户分配的频谱使用效率并不高。因此,如何提高频谱使用效率现已成为热点研究课题。认知无线电系统作为一个智能无线通信系统,在确保主用户在其授权频段上具有优先使用权,并且感知用户的接入不影响主用户性能的前提下,感知用户设备能通过自主的调整,在当前未使用的空闲频段上通信,提高频谱的使用效率。认知无线电的概念一经提出,国内外的许多学者就表现出了极大的兴趣,对其展开了广泛而深入的研究,并且获得了许多研究成果。主要包括频谱检测、认知引擎、动态频谱管理、端到端重配置等方面。其中,用于感知信道占用状态的频谱检测技术是认知无线电系统中非常重要的问题之一。频谱检测的目的有两个：一是感知用户通过切换到其他可用频段或者将自己对授权用户的干扰限制到一个可以接受的范围从而避免对授权用户造成干扰；二是,感知用户应该有效利用频谱空洞去满足系统性能要求。因此,认知无线电中的频谱感知准确度对于主用户和次用户的性能有至关重要的影响。国内外学者们从频谱能量、频谱特征、协同检测等方面对频谱检测展开研究,提出了能量检测、匹配滤波检测、循环特征检测、协作频谱检测等算法,显著提高频谱检测的精度,而且还可以减小单个认知节点的负担。然而,当前无线通信技术向移动、宽带、高速的方向不断发展,在未来的宽带无线通信系统中要支持Gb／s甚至数Gb／s的分组数据传输速率,这使得系统带宽不断增加,同时,也要求极其灵活的频谱资源分配。这就对频谱检测技术提出了新的挑战：需要在宽频带范围内对频谱进行快速、有效和动态的检测。因此研究认知无线电系统中的宽带频谱检测技术,提高频谱利用率,以适应未来高速率移动无线通系统是非常必要的。目前国内外研究人员针对宽带频谱检测提出了一些研究方法,主要包括本地单节点多窄带并行链路检测和单节点宽频带检测两种类型。单节点宽频带检测中,每个节点只需要一个射频前端,通过模拟数字转化器(Analog to Digital Converter, ADC)可以灵活地检测功率谱密度动态变化的宽带信号。然而,在宽频带范围内,根据奈奎斯特采样定律,频谱检测很难跨越超高速ADC芯片和海量存储技术的壁垒,现有的A／D器件和存储芯片难以胜任。在全频段内,如何对接收到的多个目标信号进行检测和判决,己经成为制约宽带频谱检测的一个难题。压缩感知理论为数据采集技术带来了革命性的突破,将压缩感知理论与宽带频谱检测技术相结合,可以解决ADC硬件采样频率不足的问题。基于“压缩感知”的宽带频谱检测技术是一种新型信号处理技术,其在理论上给出了可以利用频谱信号的稀疏性进行宽频带范围检测,但由于其处于研究起步阶段,大多数研究者直接把压缩感知理论套用到认知无线电中的宽带频谱检测,而经过压缩采样后,后端恢复算法的复杂度非常高,在实际检测中带来了严重的时延问题。针对宽带认知无线电系统中频谱环境动态、异构和宽频带等特点,论文以压缩感知理论为切入点,通过明确模拟信息转换器(AIC)与宽带异构频谱的内在关系,分析认知无线电中对宽带频谱信号检测的需求特点,利用源信号稀疏模型与信号先验信息的内在耦合机理,引入块稀疏信号压缩感知的理论,建立宽带认知无线电系统中的频谱动态检测模型,在主用户网络干扰容限约束条件下,研究并设计出一套适用于当前认知无线电架构的、符合特定信号频谱特征的宽带频谱检测策略。使得认知无线电技术更好地适用于未来的无线通信系统,最大限度地利用频谱资源,为认知无线电技术的进一步应用提供可选的解决方案。本文的主要创新和贡献如下： 1.提出一种基于压缩感知的非稀疏信号导频检测方法,该方法是基于压缩域中的谱估计算法,直接使用压缩感知得到的观测值进行频谱检测,从而降低数据量和算法复杂度,减小检测延时,以解决认知节点有限的计算能力与压缩感知恢复算法较高的计算复杂度之间的矛盾。该方法在压缩感知理论的框架下,以主用户的频谱划分为先验信息,设计与划分结构相匹配的导频图案,在不恢复采样信号的前提下,利用数字傅里叶变换的线性运算性质,对采样信号在压缩域进行线性运算,只保留频谱空洞处导频的频谱信息,通过后端信号处理模型,直接从压缩采样的观测值中估计非稀疏信号的参数如载波频率、带宽、功率等。通过设计仿真,验证了所提出的导频检测方法,证明其能够提高压缩感知理论对非稀疏信号检测准确性,达到减小重构复杂度、增强接收机对信号类型的鲁棒性的目的。 2.提出一种压缩感知框架下的源信号频域采样结构,它是一种基于频域采样理论的信号采样方法,通过在采样过程中加入变换基矩阵,来充分利用主用户信号在变换域中的稀疏特性,以解决现有压缩感知采样结构中没有变换基的问题,提高对频域稀疏信号的检测准确性。该方法通过对现有采样结构的研究,设计了一种带有变换基矩阵的频域随机解调器(FRD),该结构通过多路并行通道,用随机化的不同阶频域采样信号对原始信号进行预处理,使得对信号的采样可以在频域进行,将处理结果通过积分器后分别进行低速抽样判决,以获得对模拟信号的压缩测量值。通过设计仿真,与现有采样器进行对比,所提出的FRD采样结构在现有硬件设备的条件下,能够对信号进行频域采样,提高检测频域稀疏信号的准确性、减小重构复杂度、扩展前端硬件的局限性。 3.提出一种基于块压缩感知的OFDM信号检测方法,利用OFDM信号在频域的结构化特征,挖掘信号的块稀疏特性,用更具结构性的测量矩阵取代随机测量方式,对较低块稀疏度的OFDM信号进行检测,以解决用压缩感知理论对此类信号采样恢复的高复杂度和低准确性的问题。该方法通过对已知的具有特殊结构的信号的研究,根据源信号的先验信息,分析OFDM信号的结构化特征,构建源信号的块稀疏模型,设计与信号结构相匹配的测量矩阵,通过对结构化信号进行联合子空间向量的变换,以获得尽量少的测量数据。通过仿真,与传统方法相对比,所提出的检测方法提高了块稀疏信号检测准确性、减小OFDM信号的重构复杂度,将标准的稀疏度的前提条件扩展到包含更加丰富的信号类型。
[Abstract]:The cognitive radio system is one of the most important problems in cognitive radio system .

the purpose of spectrum detection is twofold : one is to sense the user to avoid interference to authorized users by switching to other available frequency bands or by limiting their interference to authorized users to an acceptable range ;
Second , the perceptual user should effectively utilize the frequency spectrum cavity to meet the system performance requirements . Therefore , the spectral perception accuracy in the cognitive radio has a crucial influence on the performance of the primary user and the secondary user .

This paper studies the spectrum detection technology from the aspects of frequency spectrum energy , frequency spectrum characteristic , cooperative detection and so on , and puts forward some algorithms such as energy detection , matching filter detection , circular feature detection and cooperative spectrum detection .

The compressed sensing theory brings revolutionary breakthrough to the data acquisition technology , combines the compression sensing theory with the broadband spectrum detection technology , and can solve the problem of insufficient sampling frequency of the ADC hardware . Based on the " compression sensing " , the broadband spectrum detection technology is a novel signal processing technique .

Aiming at the characteristics of spectrum environment dynamics , heterogeneous and broadband in the broadband cognitive radio system , the paper takes the compression - sensing theory as the entry point , analyzes the inherent relationship between the analog - to - analog information converter and the wide - band heterogeneous spectrum , analyzes the inherent coupling mechanism of the source signal sparse model and the signal - prior information , and introduces a broad - band spectrum detection strategy applicable to the current cognitive radio architecture and conforms to the characteristic of the spectrum of the specific signal .

The main innovations and contributions of this article are as follows :

The invention provides a non - sparse signal pilot detection method based on compression perception , which is based on a spectral estimation algorithm in a compressed domain , directly uses the observation value obtained by the compression perception to carry out frequency spectrum detection , thereby reducing the data amount and the algorithm complexity and reducing the detection time delay so as to solve the contradiction between the limited computing capacity of the cognitive node and the computational complexity of the compression perception recovery algorithm .

In the framework of the compression perception theory , the frequency spectrum of the main user is divided into a priori information , and a pilot pattern matched with the division structure is designed .

The invention provides a source signal frequency domain sampling structure under a compression sensing framework , which is a signal sampling method based on a frequency domain sampling theory .

A frequency - domain random demodulator ( FRD ) with transform matrix is designed by the research of the existing sampling structure . The structure is pre - processed by means of multi - channel parallel channel , and the original signal is pre - processed by randomizing different order frequency domain sampling signal , so that the sampling of the signal can be carried out in the frequency domain , and the result is compared with the existing sampler . The proposed FRD sampling structure can sample the signal in frequency domain , improve the accuracy of the sparse signal in frequency domain , reduce the reconstruction complexity and expand the limitation of the front end hardware .

3 . An OFDM signal detection method based on block compression perception is presented , which uses the structured feature of the OFDM signal in the frequency domain , the block sparse characteristic of the mining signal , the random measurement mode is replaced with a more structured measurement matrix , and the OFDM signal with lower block sparsity is detected to solve the problem of high complexity and low accuracy of the sampling recovery of the signal by using the compression perception theory .

the method improves the detection accuracy of the block sparse signal , reduces the reconstruction complexity of the OFDM signal , and extends the precondition of the standard sparsity to a signal type containing more abundant signal .
【学位授予单位】：北京邮电大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN92

【参考文献】