高区分度欠采样宽带谱分析器设计  

Design of High-Discrimination Undersampling Wideband Spectrum Analyzer

作　　者：黄翔东 宋金水 李燕平 杨自凯 Huang Xiangdong;Song Jinshui;Li Yanping;Yang Zikai(School of Electrical and Information Engineering,Tianjin University,Tianjin 300072,China;School of Artificial Intelligence,Heibei University of Technology,Tianjin 300401,China)

机构地区：[1]天津大学电气自动化与信息工程学院,天津300072 [2]河北工业大学人工智能与数据科学学院,天津300401

出　　处：《天津大学学报（自然科学与工程技术版）》2024年第8期847-855,共9页Journal of Tianjin University：Science and Technology

基　　金：青海省基础研究计划面上项目(2021-ZJ-910).

摘　　要：为克服现有基于多陪集、调制宽带转化器等欠采样结构的谱分析器存在谱成分区分度差、欠采样通道数多、有效探测带宽窄的严重缺陷,提出基于全相位滤波器对半分解的互素欠采样谱分析器设计.首先,剖析了经典互素谱分析存在伪峰效应的原因.然后,具体提出基于最小尺寸的全相位滤波器对半分解的解析方法设计互素谱原型滤波器,并从中衍生出两路互素谱分析结构并行的谱分析器,从而提高了谱区分度;分析了两路并行互素谱输出的频点位置关系,通过对两路输出做交叉合并,生成双倍频谱分辨率的全景谱,既可消除伪峰,又无需增加欠采样通道数(即为最小值2);在全景谱重构计算中,通过将上、下通道支路的互相关扫描代替现有欠采样谱分析的压缩感知求解计算,绕过了对信源带宽稀疏度的要求,从而不仅拓宽有效探测带宽,还获得了密集谱区分能力.最后分别以多子带的数字已调信号、线性调频信号为信号模型,对所提出的高区分度宽带谱分析器的谱估计效果进行验证.仿真结果表明,该谱分析器可以消除现有互素谱分析方法存在的伪峰效应,且可在总体欠采样率不足0.5倍调制宽带转化器欠采样率的条件下,可区分的谱成分数目提升至其2.5倍以上.在跨频段、超分辨率的欠采样频谱感知中具有广阔的应用前景.The mainstream undersampling spectral analyzers(multicoset sampling based and modulated wideband converter(MWC)based,etc.)are limited by the following deficiencies:low spectral component discrimination,excessive consumption of undersampling channels and narrow effective detection bandwidth.To overcome these defi-ciencies,this paper proposes an undersampling wideband spectrum analyzer based on the half-decomposition of allphase filters.Firstly,the reasons for the spurious peaks in the classical coprime spectral analysis are anatomized.Then,the enhancement of spectral discrimination arises from the combination of minimum-sized half-decomposition of all-phase filters and its extension of two paralleled coprime analyzers.Moreover,the frequency points position relationship of the outputs of the two parallel coprime analyzers is analyzed.The cross-merging of the outputs of these two analyzers can generate the panoramic spectrum with double spectral resolution,which not only eliminates spurious peaks,but also helps to fix the two undersampling channels.Besides,the proposed analyzer reconstructs the panoramic spectrum via implementing cross-correlation operation on the datastreams of top and bottom paths rather than via the compressive compression approaches,which actually bypasses the requirement of source sparsity.As a result,the proposed analyzer not only broadens the effective detection bandwidth,but also acquires the ability of discriminating dense spectra.Finally,the spectrum estimation performance of the proposed spectrum analyzer is validated using digital modulated signals and linear frequency modulated signals with multiple subbands as signal models.Numerical results showed that,the proposed analyzer not only can eliminates spurious peaks in existing co-prime spectrum analysis but also can identify 2.5 times of spectral components as the MWC based analyzer does,under the condition that the former only consumes less than half of samples of the latter one.The proposed spectrum analyzer has broad prospects in cros

关 键 词：高区分度 欠采样 互素谱分析 调制宽带转化器 全相位滤波器 

分 类 号：TN911.7[电子电信—通信与信息系统]