基于张量分解的光子学微波信号频域空域联合压缩感知研究

The multidimensional parameter sensing of microwave signals is the basic function and key technology in the field of electronics. The sensing via photonic techniques has significant advantages, such as large instantaneous bandwidth, low frequency independent loss, and strong electromagnetic interference immunity. Under this background, photonics-assisted compressed sensing in joint frequency-spatial domain for microwave signals wide spread in frequency and space via tensor decomposition is explored. The key parameters, frequency and direction-of-arrival, are jointly estimated, by exploiting the sparsity and multilinearity within the signals. Firstly, microwave photonic antenna array is constructed with spatial sparsity and multilinearity, based on which joint spatial time compressed sampling is realized under the Tucker decomposition of a tensor. Secondly, the multidimensional compressed sensing recovery algorithm based on the low rank characteristic is designed to reconstruct the full signal tensor. The key parameters, frequency and direction of arrival, are then jointly estimated from the PARAFAC decomposition of the full signal tensor. Finally, the resolution, receiving gain, and stability of the implementation are optimized and improved by jointly debugging both the software and hardware. The benefit from the introduction of tensor decomposition is twofold: first, the receiving channels are compressed, so the sampling and transferring data is significantly reduced; second, the model identifiability is boosted, as a result, the multidimensional parameters of each signal are estimated more accurately and automatically paired. These research work and results have both positive theoretical significance for the detection and analysis of broadband microwave signals, and important application value in the rail transit and the civil-military inosculation fields.

微波信号多维感知是电子信息领域的基础功能和关键技术，并且以光子技术实施感知具有瞬时带宽大、频率相关性损耗小、抗电磁干扰强等显著优势。在此背景下，围绕微波频率和到达角这两维核心参数的联合感知需求，本项目针对微波信号蕴含的稀疏性和多线性，探索基于张量分解的“频域宽开”和“空域宽开”光子学联合压缩感知机制及解决方案。首先，从空域稀疏布阵和多线性出发构建微波光子天线阵列，在张量Tucker分解机制下实现空时压缩采样。进而，设计基于张量低秩特性的多维压缩恢复算法，重建原始张量信号，再融合张量PARAFAC分解联合估计频率和到达角。最后，结合软硬件联合调试，优化并提高实施方案的分辨率、接收增益和稳定性。张量分解的引入显著压缩接收通道、降低采集和传输数据量，同时增强模型辨识性，并自动配对多维参数。这些研究工作及成果对检测、分析宽频微波信号具有重要理论意义，并在轨道交通、军民融合领域具有重要应用价值。

微波信号多维感知是电子信息领域的基础功能和关键技术，在移动通信系统技术和卫星通信技术有重要应用，并且以光子技术实施感知具有瞬时带宽大、频率相关性损耗小、抗电磁干扰强等显著优势。在此背景下，围绕微波频率和到达角这两维核心参数的联合感知需求，本项目针对微波信号蕴含的稀疏性和多线性，探索基于张量建模和分解的“频域宽开”和“空域宽开”光子学联合压缩感知机制及解决方案。首先，从空域稀疏布阵和多线性出发构建微波光子天线阵列，在张量Tucker分解机制下实现空时压缩采样。2020年度开展了基于张量分解的光子学微波信号频域空域联合压缩感知研究平台搭建，构建微波光子信号空频两维压缩感知传输链路测试平台；同时开展了一维稀疏布阵天线阵列的布置研究。随后，设计基于张量低秩特性的多维压缩恢复算法，重建原始张量信号，再融合张量PARAFAC分解联合估计频率和到达角。2021年度在平台基础上开展了多通道空频联合压缩感知平台信号处理研究工作，系统联试、通道校准、和基带处理开发框架构建；进一步完善了稀疏天线阵列布置几何特征辅助相位处理的统一理论；针对光子学辅助的阵列信号处理，提出光相控阵Blass矩阵的低耗散功率综合方法，与现有方法相比，耗散功率降低了99.63%。最后，结合软硬件联合调试，优化并提高实施方案的分辨率、接收增益和稳定性。张量分解的引入显著压缩接收通道、降低采集和传输数据量，同时增强模型辨识性，并自动配对多维参数。2022年度，设计并验证了光子学微波信号频域空域联合压缩感知系统框架，对三个5GHz范围内不同空间来向的信号进行分辨，并实现参数自动配对。这些研究工作及成果对检测、分析宽频微波信号具有重要理论意义，随着高性能光子器件及其阵列的成熟，项目成果未来在轨道交通、军民融合领域的移动通信系统技术和卫星通信技术中具有重要应用价值。

内容获取失败，请点击重试