基于压缩感知的通信信号处理理论研究

The purpose of this research project is to explore new compressed sensing theories and techniques for sub-Nyquist sampling and analysis of complex wideband communication signals. Our research consists of the following four aspects. Firstly, we aim to seek a basis on which the complex wideband communication signal or its higher-order statistics has a sparse or approximate sparse representation. Secondly, we develop new compressed signal processing theories and techniques to accomplish the detection, estimation, and modulation recognition of wideband communication signals based solely on the down-sampled measurements, which can help overcome the difficulty of Nyquist sampling of wideband signals. Thirdly, by exploiting the block-sparse structure of wideband communication signals, we develop sparse signal recovery algorithms which have very low computational complexity but meanwhile can still provide state-of-the-art recovery performance. Lastly, one-bit compressed sensing techniques are developed to dramatically reduce the hardware complexity. The ultimate goal of this research is to overcome the deficiencies in existing compressed sensing techniques, and develop efficient compressed sensing methods for sub-Nyquist sampling and analysis (such as estimation and modulation recognition) of complex wideband communication signals.

本项目针对大带宽、多信号、多调制样式的复杂通信信号处理面临的奈奎斯特采样和信号特征提取等关键技术问题，开展基于压缩感知的通信信号处理理论研究。具体的来说，研究通信信号的稀疏表征，设计有效的字典学习算法，寻求通信信号或其特征量的稀疏表征基；研究压缩信号处理理论与算法，利用压缩数据完成对通信信号的检测、定阶、和调制识别等问题，克服奈奎斯特采样瓶颈问题；针对宽带通信信号的结构稀疏特性，研究新型高精度、低复杂度稀疏重构算法，满足宽带信号重构的实时性和高精度要求；研究基于一比特压缩感知和压缩信号处理方法，降低硬件设计复杂度，减少采样数据量，更高效的完成数据分析和处理。本项目的研究，有助于压缩感知方法的完善，促进解决大宽带复杂通信信号处理的瓶颈技术问题，促进无线通信和军事监测事业的发展。

本项目针对多频带信号(稀疏多带信号)开展了基于压缩感知的信号处理理论与方法研究，目的是通过利用压缩感知的理论与方法，突破对大带宽多频带信号的奈奎斯特采样瓶颈的限制，以次奈奎斯特采样频率完成对大带宽多频带信号的侦察。具体来说，本项目开展的研究工作如下：首先，通过利用多频带信号在频域的稀疏性，将信号重构问题转换为压缩感知问题，并提出了无需矩阵求逆的快速稀疏贝叶斯信号重构方法；另外，针对多频带通信信号的结构稀疏特性，提出了低复杂度的基于结构耦合的块状稀疏贝叶斯信号重构算法，满足信号重构的实时性和高精度要求；此外，在压缩采样的框架下，提出了一种快速的功率谱重构方法，可以在无需重构原始信号的情况下对宽带频谱进行检测，为实时宽带频谱感知提供了一种实用的解决方案；最后，研究了基于一比特量化数据的压缩感知方法，分析了量化门限对信号重构性能的影响，并在此提出上提出了自适应量化的压缩感知方法，大幅度提升了基于一比特量化数据的信号重构性能。相关研究成果已申报国家发明专利4项，在IEEE期刊发表论文8篇。