【作者】 刘建；

【导师】 王勇； 苏健；

【作者基本信息】 西安电子科技大学 ， 工程硕士（专业学位）， 2019， 硕士

【摘要】 近几年5G等通信技术快速发展的同时,无线用户数在连续增加,造成频谱资源变得愈发紧张。由于现有的频谱分配方式为静态分配,主用户很有可能在相当长的一段时间内占用授权频段却未进行通信,从而使得频谱资源利用率很低。为了使频谱资源的利用率得以提高,认知无线电技术被提出,而对所关心的授权频段进行准确的频谱感知是认知无线电技术的前提。通过对周围环境进行频谱感知,能够保证在所关心频段未被主用户使用时,认知用户能快速安全地接入该频段进行通信,因此频谱资源得到充分利用。本文首先以经典算法中的协方差矩阵频谱感知算法在多协作用户系统下为例进行改进研究。在多协作用户系统中,若协作用户数目少,该类算法的检测性能会受到很大影响,因而将“矩阵拆分重组”的方法引入该类算法,分析了CAV(Convariance Absolute Value)算法以及MME(Maximum Minimum Eigenvalue)算法采用“矩阵拆分重组”后的性能变化,给出了性能仿真。通过研究拆分重组的不足,进而提出了“矩阵滑窗重组”的方法,将该方法应用到CAV算法以及MME算法,通过理论结合仿真,更进一步提高了协作用户数少时的检测性能。然后本文对已有的基于角度维度的WMUSIC(Weighted Multiple Signal Classification)算法的盲频谱感知算法进行改进研究。空间角度维度的频谱感知算法通过调整天线角度,使得认知用户能在主用户占用频谱资源时仍然可以进行通信,实现同时同频不同角度的资源利用。因此在信号检测阶段,通过对空间谱构造多级检测统计量,并给出每一级检测统计量对应门限值的求取方法,结合仿真验证了检测性能的提高。在角度估计阶段,通过分析阵元间距与角度分辨率的关系,结合实际工程应用,给出了无虚假谱峰的最大阵元间距,最大阵元间距大于载波半波长。理论分析结合仿真共同验证了此时的阵元间距对应的角度分辨率更高;现有的WMUSIC盲频谱感知算法无法处理相关信号,对此进行改进,结合Matlab仿真验证,使得改进后的算法能区分相关的主用户信号;考虑多普勒频移,给出了多普勒频移下的WMUSIC盲频谱感知算法的分析。最后,对已有的TCT(Two-sided Correlation Transform)算法进行改进,改进后既无需角度预估计也不需要信源数目估计,从空间角度维度提出了一种基于TCT算法的宽带信号盲频谱感知算法。宽带信号检测阶段,从空间谱构造检测统计量,并给出检测统计量对应的门限值的选取方法,通过检测统计量与门限值的比较来感知宽带信号是否存在,并通过Matlab仿真,调整不同参数的变化,给出了该频谱感知算法的参数变化对于检测性能的不同影响。而在宽带信号角度估计阶段,构造新变量替代聚焦矩阵,一定程度上提高了角度估计的准确性。估计出主用户宽带信号的角度后,通过分析自适应波束技术,选取LCMV(Linear Constrained Minimum Variance)准则,使得主用户与认知用户互不影响的进行各自的通信。更多还原

【Abstract】 In recent years,the number of wireless users is increasing continuously with the rapid development of 5G and other communication technologies,which makes the spectrum resources become more and more tense.Because the existing spectrum allocation method is static,it is possible for the primary user to occupy the authorized frequency band for a considerable period of time without communicating,which makes the utilization of spectrum resources very low.In order to improve the availability of spectrum resources,cognitive radio has been proposed.Accurate spectrum sensing of the authorized frequency bands concerned is the premise of cognitive radio technology.By spectrum sensing of the surrounding environment,it ensures that cognitive users can access the band quickly and safely when it is not used by the primary users,so the spectrum resources are fully utilized.First,taking the covariance matrix spectrum sensing algorithm of classical algorithm in multi-user system as an example,we make the performance improved.In a multi-user system,if the number of collaborative users is small,the detection performance can be greatly affected.Therefore,the method of "matrix splitting and reorganization" is applied on this kind of algorithm.The performance changes of CAV(Convariance Absolute Value)and MME(Maximum Minimum Eigenvalue)after this method are analyzed,and the performance simulation is given.We study the disadvantage of splitting and reorganization,and then put forward a method of "matrix sliding window reorganization" applied on CAV and MME.Combining theory with simulation,it further improves the detection performance when the number of collaborative users is small.We then improve the WMUSIC(Weighted Multiple Signal Classification)algorithm based on angle.the spectrum sensing algorithm of spatial angle dimension enables cognitive users to communicate when primary users occupy spectrum resources by adjusting the antenna angle.With different angles,it realizes the resource utilization of the same frequency at the same time.In the signal detection stage,multi-level detection statistics are constructed by spatial spectrum,and the corresponding threshold value of each level detection statistics is given.Combined with simulation,we verify the improvement of detection performance.At the angle estimation stage,by analyzing the relationship between element space and angular resolution,and combing practical engineering application,the maximum element space without false spectral peaks is given,which breaks through the limitation of half-wavelength element spacing.The improvement of angular resolution is verified by simulation.WMUSIC blind spectrum sensing algorithm can not process related signals,so we do some improvement on it.Combined with Matlab simulation,the improved algorithm can distinguish related primary user signals.Besides,considering Doppler frequency shift,the analysis of WMUSIC blind spectrum sensing algorithm under Doppler frequency shift is given.Finally,the existing TCT(Two-sided Correlation Transform)algorithm is improved.Neither angle estimation nor source number estimation is needed.A blind spectrum sensing algorithm from the spatial dimension for wideband signals based on TCT algorithm is proposed.In the phase of wideband signal detection,detection statistics are constructed from spatial spectrum,and the method of deciding threshold values corresponding to detection statistics is given.The existence of wideband signals is perceived by comparing detection statistics with threshold values.Through Matlab simulation,the different effects of the parameters of the spectrum sensing algorithm on the detection performance are given.At the wideband signal angle estimation stage,we construct a new variable instead of focusing matrix to improve the accuracy of angle estimation to a certain extent.After estimating the angle of the wideband signal of the primary user,the LCMV(Linear Constrained Minimum Variance)criterion is selected by analyzing the adaptive beam technology,so that the primary user and the sensing user can communicate separately without affecting each other.更多还原