水声信道平稳/时变混合多径模型及估计方法研究

Due to significant impacts on sound propagation from complicated multi-path spread and time variations of underwater acoustic (UWA) channels, accurate, efficient and robust estimation of UWA channels is important basis and frontier in research and design of UWA communication and sonar systems. Considering time-varying and structural features of UWA channels, the present research will focus on the following aspects..Firstly, according to different dynamic features caused by sound propagation mechanism, a novel hybrid stationary/time-varying multipath model is proposed to enable the exploitation of the sparisity as well as dynamic features of UWA channels to improve the structural-dynamic match with practical cases. The effectiveness and channel adaptability of the proposed model will be comprehensively verified. .Secondly, the estimation methods that can adapt to the features of stationary components as well as time varying components will be developed respectively. Specifically, under the model decomposition, decreasing dimension of parameter space associated with time varying components makes it possible to exploit the excellent performance of support vector machine (SVM) to develop fast convergence, low complexity estimation algorithm for time varying multipath components. .Finally a high efficiency, robust UWA channel estimation method based on the hybrid stationary/time-varying multipath model will be derived. During the proposed research, the innovative progresses in theoretical analysis and practical technology will offer novel ways for the high performance estimation and feature exploitation of UWA channels, which have considerable potential for applications in the research of advanced UWA communication and sonar.

由于海洋水声信道复杂的多径扩展和时变性对声传播带来的严重影响，准确、高效地获取信道特性是分析和设计水声通信、声纳系统的基础和重要研究前沿。针对水声信道中多径的时变及结构分布特点，本项目将围绕以下两个方面展开研究。.首先，根据声传播机理导致的信道多径动态特性不同，提出平稳/时变混合多径模型进行多径分量的分解，研究通过多径信道稀疏、时变特征的高效表征提高模型的结构和动态适配性，并对该模型描述水声信道的有效性进行充分验证；同时，研究模型中平稳、时变多径分量的针对性估计方法，特别是基于模型分解后分量解空间维数的下降，探索采用支持向量机的小样本动态学习能力解决时变分量的快速、低复杂度估计。在此基础上设计高效、稳健的水声信道平稳/时变混合多径估计方法。.通过上述研究中取得理论和关键技术的创新和突破，探索为水声信道特性的高效获取与匹配开辟新的途径，对高性能水声通信及探测系统的研究有积极意义。

随着海洋信息获取与传输的需求日益增加，水声通信技术已成为海洋高科技的前沿和热点。然而，声音在海洋中传播因发生反射，衍射，甚至扩散等现象，容易导致水声通信过程中产生复杂的多径效应，即产生接收端的信号混叠，从而恶化通信效果。因水声信道本身所含的复杂的物理特性造成了水下的通信困难，进一步制约着水声通信的高速，大容量和有效带宽利用等技术的研究和发展。研究表明，利用水声信道的稀疏特性在压缩感知框架下进行信道估计可改善性能，同时可以减少因噪声引起的干扰。.但是，由于呈现与经典无线信道完全不同的稀疏特性，水声信道多径反射、散射等复杂传播导致的不同类型、混合（平稳/时变、离散/散射）的稀疏特性对目前被广泛采用的稀疏信道估计方法造成严重问题。项目基于水声信道呈现的复杂稀疏特性，在传统稀疏估计l1、l0范数约束的基础上，提出近似p范数及非均匀范数进行稀疏信道建模，通过引入调整机制从近似范数角度对不同类型、混合的稀疏特性进行自适应；同时，提出了梯度稀疏恢复结合梯度解投影的迭代求解方法，并对求解方法的收敛性进行了理论分析。.不同类型稀疏信道下的数值仿真以及不同海域进行的海试实验均验证了项目所提近似范数信道稀疏模型及参数迭代求解方法的有效性，该方法进一步结合信道匹配水声通信机和有效改善不同多径类型下的水声通信性能。.项目研究成果在本领域重要SCI刊物发表并被引用40余次，得到国内外同行的关注。项目组基于本项目成果研发的水声通信机可适应不同海洋环境下的不同多径，目前已应用于海洋工程领域。