基于压缩传感理论的宽容匹配场处理技术研究

Sonar technology is faced with the problem of development for detection and estimation theory and method in uncertain oceanic environments. From the free field to the waveguide and toward dynamic uncertain environments, the operating environment of sonar signal becomes more and more complex, but provides also more and more knowledge, which supports to estimate the only bearing to bearing-range-depth. However, the increasing complexity brings more uncertainty, thereby the mismatch problem is becoming more and more serious, robust detection and estimation theories and methods are encouraged to be developed.Luckily,the stuctural knowlege is always low deimensional and roubust. This project is aimed at the problems mentioned above, makes full use of stuctures of acoustic signal - - both of sparsity: sparsity of channel transmission, the spatial spectrum and waveform and the propagation knowledge of oceanic channel, to solve some key problems in the robust detection and estimation of sonar. The problems are matched field processing of short array and sparse array, robust matched field processing due to uncertainty and variability of environmental parameters and not exactly known system parameters, which in turn lead to mismatch of sparse matrix, compressed sensing Bayesian matched field processing of moving target based on state-space model, and fast algorithm with utilizing of block sparsity knowledge. On this basis, a real-time simulation and test system will be exploited to operate in an appropriate scale of processing platform, and to be verified by the data sampled in sea test. We will try our best to both promotion of the theory of compressed sensing matched field processing and laying the foundation for its forward to practical applications.

声纳技术面临着发展不确实海洋环境下探测理论和方法问题。从自由场到波导再朝向动态不确实性环境，声纳信号运行的环境越来越复杂，提供的知识也越来越多，由只方位估计到方位-距离-深度估计，但复杂性提高的同时带来的是不确实性增大，失配问题日趋严重，宽容性处理的理论和方法急待发展。本项目针对上述问题，充分发掘利用结构性- - 信道传输、空间谱的稀疏性和信道传播知识，解决声纳宽容性探测中的若干关键问题。包括短阵和稀疏布阵的压缩传感匹配场处理(CSMFP)、参量不确知和变化导致稀疏基矩阵失配时宽容CSMFP、运动目标基于状态-空间模型的贝叶斯CSMFP，以及基于块稀疏性知识的快速算法等。开发实时软件仿真和测试系统，利用实测数据进行验证。为CSMFP的理论发展和实用化打下基础。

我们把水下航行的物体、海洋动态信道等都可以看作为广义的水下目标，对水下目标三维定位一直以来是水声信号处理的难点之一。因为海洋范围大，空间与时间存在变化性，海洋环境的知识难以精确已知，从而导致基于声传播模型的水下目标三维定位多数情况下不能正确定位。项目研究开发的压缩传感匹配场处理方法，把海洋中声传播看作为随机过程，由统计知识构建观测方程，具有较好的宽容性，同时，再进一步利用目标数目相较匹配场处理搜索网格的稀疏性知识，通过仿真和实际数据分析，理论限与性能限的计算，对CS-R方法与常规匹配场处理方法、最小方差不失真响应、1-SVD压缩传感方法进行了多方面的比较，结果说明CS-R方法能够在宽容的基础上实现高分辨，即具有常规方法的宽容性，又有着压缩传感方法的高分辨性，是一种有应用前景的方法。CS-R将应用于在目前课题组承担的海军预研项目，用于水下目标的三维定位。. 海洋环境也是一个弥漫、复杂的动力学系统，在其中除了集中性目标，还有海洋地球物理波这一类弥漫性目标。声波是海洋中能够长距离传输的波，声波在海洋中的传播受到海洋介质特性和海洋动力学性质的影响，例如声速随深度的函数导致声线多路径传播，不同路径以不同的时延和到达角从声源到达接收，导致接收的波形发生扩展。在水声通信与海洋环境参量估计时，我们常常需要对不同路径的信道进行估计与跟踪，本项目开发研究的压缩传感-卡尔曼滤波（CS-KF）方法可实现动态海洋信道的估计与跟踪，可应用水声通信和海洋监测中。课题组目前正承担重点研发专项，需要实现大范围长时间的海洋声速场和流速场的估计，CS-KF方法可应用于深海多路径的时延和到达角估计与跟踪，进而通过逆问题求解实现海洋声速场和流速场的估计。