基于压缩感知的超宽带通信信号处理与检测机制研究

Analog-to-digital conversion is the bottleneck hampering the development of ultra-wideband communications. Based on the concept of software-defined radio (SDR), the proposed research utilizes the emerging compressed sensing theory to lower the sampling rate of ultra-wideband signals and attempts to realize the signal processing functions and symbol detection of ultra-wideband receivers in compressed digital domain. Firstly, starting from the statistics of ultra-wideband multipath signals and utilizing principal component analysis (PCA) and/or independent component analysis (ICA), we will investigate the sparse decomposition of ultra-wideband signals and construct statistically optimal dictionary of sparse bases such that both the measurement circuitry complexity and the reconstruction algorithm complexity can be reduced effectively. Based on the idea of "matched-filtering", we will investigate the design methodology of preprocessing matrix based on which compound measurement matrices with efficient signal energy capture and ability of noise suppression can be constructed. After that, based on the measurement sequences obtained, we will investigate the signal processing and symbol detection mechanism. Efficient training sequence-assisted synchronization algorithm will be established and algorithm for narrowband interference detection and suppression will be proposed also. Two symbol detection mechanisms, named measurement subspace detection and signal subspace detection, will be studied. For each symbol detection mechanism, channel estimation algorithm, diversity combining algorithm and symbol detection algorithm will be proposed. This research will meet the urgent needs of the development of internet of things (IOT) and also provide new ideas and theoretical fundamentals to the future wideband wireless communications.

A/D转换是制约超宽带通信技术发展的技术瓶颈。本项目基于软件无线电理念，利用新兴的压缩感知理论实现超宽带信号低速率采样，并试图在压缩数字域实现超宽带接收机的信号处理与检测功能。首先，从超宽带多径信号统计特征入手，利用主分量分析、独立分量分析研究超宽带信号稀疏表示方法，构建统计意义下最优的稀疏基字典，降低观测电路实现复杂度和重构算法复杂度；基于"匹配滤波"思想研究预处理矩阵设计方法，构建信号能量捕获效率高、噪声抑制能力强的复合观测矩阵。然后，基于压缩感知序列，研究压缩数字域信号处理与检测机制，提出高效的训练序列辅助的同步捕获算法和窄带干扰检测与抑制算法，探讨"观测子空间检测"和"信号子空间检测"两种符号检测机制，针对两种检测机制分别提出可行的信道估计算法、分集合并算法和符号检测算法。本项目的研究将适应物联网发展的迫切需求，并可为未来宽带无线通信技术发展提供新思路和理论依据。

超宽带无线通信是最具竞争力的物联网应用支撑技术，为突破传统采样理论技术瓶颈，本项目基于软件无线电理念，研究基于压缩感知的低采样率超宽带数字接收机技术。在国家自然科学基金资助下，在以下三个方面开展了研究工作并取得了一系列研究成果：1.从超宽带多径信号统计特征入手，研究了超宽带信号稀疏表示方法。基于主分量分析理论，提出变换域信号逼近模型和基于特征向量的稀疏字典构建方法，基于移不变模型，提出了基于导数波形和希尔伯特变换波形的多核稀疏字典构建方法，构建的稀疏字典使超宽带信号更为稀疏，可以有效降低观测电路实现复杂度和重构算法复杂度；2. 基于“匹配滤波”思想研究了预处理矩阵设计方法，与随机观测矩阵融合生成复合观测矩阵。提出了基于超宽带信号能量分布先验知识的加窗方法和基于超宽带信号主分量统计分布特征的预处理矩阵设计方法，提高了观测矩阵的信号能量捕获效率和噪声抑制能力。3. 研究了压缩数字域超宽带通信信号处理与检测机制。探讨了“观测子空间检测”和“信号子空间检测”两种符号检测机制，针对两种检测机制，提出了可行的信道估计算法、窄带干扰估计与抑制方法、分集合并和符号检测算法，为实现基于软件无线电理念的超宽带数字接收机奠定了基础。本项目的研究为未来宽带无线通信技术发展和应用提供了新的思路。