宽带时变水声信道的建模与参数估计

Unlike the time-varying terrestrial radio channels, the underwater acoustic (UWA) channel is a wideband time-varying one. The Doppler effects caused by time-varying characteristic cannot simply be represented and compensated by frequency shifts, because UWA channel can cause different frequencies to be shifted in time by different amounts. This feature should be characterized by the Doppler scale spread. It is just this Doppler scale spread that severely limits the receiver performance in UWA communications. Currently, researches on the physical layer techniques in UWA communications are mostly concentrated in how to improve the communication capacity and reliability of the underwater acoustic broadband channels with time-frequency double dispersion. In this subject, we will proceed from the time-varying property, wideband nature and sparse characteristic of the underwater acoustic channel, to build a matched underwater acoustic channel model. Novel channel estimation algorithms will be studied by taking advantages of Doppler compensation technology, wavelet transform and fractional Fourier transform. Simultaneously, the sparsity of the wideband spreading function in the delay-scale plane will be exploited to model the underwater acoustic channel estimation problem within the compressed sensing framework. The compressed sensing reconstruction algorithms can be used to accurately estimate the channel parameters, including amplitude, delay and scale, from a small number of pilots or relatively short training sequence, so as to increase the spectrum efficiency. The research results are expected to perform important roles in the development of high-speed underwater acoustic communication system with independent intellectual property rights.

与陆上时变信道不同，水声信道是一个宽带时变信道。时变引起的多谱效应不能简单地用频率偏移来表示和补偿，其频率偏移量会随频率大小和时间而变化，这一特点需要使用多普勒尺度扩展来描述。正是这种多谱勒尺度扩展严重限制了水声通信接收机的性能。当前，水声通信物理层技术的研究主要集中在提高这种时频双色散宽带信道的容量和传输可靠性。本课题从水声信道的时变特性、宽带本质以及稀疏特性出发，建立与之匹配的信道模型，利用多普勒尺度扩展补偿技术，以及小波变换、分数阶傅里叶变换等多种时频分析工具研究新型信道估计算法。同时，利用宽带扩散函数在时延-尺度平面的稀疏性，建立压缩感知理论框架下的稀疏-宽带时变水声信道估计问题模型，应用稀疏重构方法对幅度、时延、多普勒尺度等信道参数进行估计，以少量的导频信号或较短的训练序列在提高系统频谱利用率的同时实现参数的准确估计。研究成果有望为研制具有自主知识产权的高速水声网络发挥重要作用。

海洋强国与智慧海洋战略的开启，对高速可靠水声通信存在广泛需求。当前水声通信亟待解决两个关键问题：一是如何在低信噪比与有限带宽下实现高速通信，另一个是如何在宽带时变信道下实现可靠传输。本研究针对这两个关键问题，主要开展了宽带水声信道的建模、高精度水声信道估计与均衡算法以及高谱效调制解调技术等的研究，同时在水声定位及水声网络方面也进行了相应的研究，以提高整个通信网络的传输速率，并搭建了实际的水声通信网络平台以进行仿真与实验验证。本研究提出了一种基于分数阶Fourier变换的多尺度多时延水声信道参数估计方法。该方法能够以迭代的方式对多分量线性调频信号按照能量大小顺序进行分离，在每次迭代中，通过一个可以校正最佳变换阶数的子迭代过程来获得尺度、时延等参数的精确估计。仿真实验表明，该方法对宽带水声信道的估计误差明显优于现有的其它方法。本研究提出多个关于宽带时变水声信道的迭代估计与均衡的高性能算法，在仿真实验中获得良好的性能。本研究还探讨了水声OFDM关键技术、OFDM子载波序号调制、单载波频域导频复用以及非正交多载波和非正交多址等高谱效调制解调技术。下一步工作将探讨如何降低本研究所提出的宽带水声信道估计与均衡算法的复杂度、以及如何将多种高谱效调制解调方案结合并应用到实际的水声通信系统中。项目项目成果有望在未来的水声通信设备中进行应用，以提高实际水声产品的通信速率或稳健性。