【作者】 周天；

【导师】 徐文；

【作者基本信息】 浙江大学 ， 信息与通信工程， 2015， 硕士

【摘要】 以声波作为载体的通信技术是实现水下远距离高速信息传输的唯一有效手段。为了提高频谱利用率,应对越来越多的高速水声通信的应用要求,相位调制技术已经取代了过去的频率调制技术成为了水下通信研究的主要内容。考虑到海洋动态性带来的信道相位响应的快速起伏变化,相位调制技术十分容易受到声传播环境的影响。总结其影响主要体现在由界面反射、介质散射带来的多径时延扩展,由界面、介质、平台运动带来的多普勒扩展,以及这些扩展随时间的变化。这样的信道通常总结为时变双扩展信道,经过该信道传播的信号会在时延和多普勒两维上展宽,并且伴随有随机起伏。当多普勒扩展较大时,无法用单一的多普勒因子进行补偿,需要联合估计出时延和多普勒两维信息,才能恢复出在信道中经过两维展宽的信号；而当信道的时变性较明显时,需要实时更新信道估计的结果,才能保证接收机性能的稳定。本文针对浅海起伏环境下水声信道具有的时延、多普勒双扩展特性及时变特性,以时变水声信道建模-双扩展信道估计-双扩展信道跟踪为主轴,研究相干水声通信技术。水声信道特性的研究是设计和仿真分析水声通信系统的基础。本文首先从理论上分析了传统的简正波模型和射线模型实现双扩展建模的方法；其次讨论了用有限的参数描述通信信道输入输出关系的路径参数化模型；然后详细研究了在已知接收发射源位置以及波导环境信息的情况下,基于镜像法模拟满足水声物理传播衰减和海面海底反射条件环境约束的双扩展信道的方法,并且对各路径的衰减幅度进行自回归建模,得到时变的水声双扩展信道模型。研究信道的时延和多普勒两维特性既可以获取收发双方和环境的空间位置信息,也可以获取平台的运动信息和环境的动态演化知识,因此研究双扩展信道能够有效、完备地利用蕴含在信号中的信息。需要指出的是双扩展信道估计问题本身是一个欠定逆问题,并不能通过传统的估计方法获得,本文从约束优化和匹配滤波两个角度来解决估计问题。从约束优化角度,相关研究已经成功地将压缩感知理论中的稀疏重构算法应用到扩展函数的估计问题,本文选择其中的正交匹配追踪算法进行理论和仿真研究。从匹配滤波的角度,本文提出了基于互模糊度函数的信道估计方法。基本原理是：计算接收信号与发射信号的互模糊度函数,得到有模糊尖峰的二维平面；并在此平面上以噪声门限估计时延,以时延维的极大值估计多普勒。论文通过仿真比较发现：正交匹配追踪算法用较多的训练序列可以得到更为准确的估计结果,但当训练序列不足时,其表现出了对算法参数选择的敏感性；基于互模糊度函数的算法则能够根据接收信噪比对算法参数进行调整,并且考虑其较低的计算复杂度,可以通过更快的更新估计来提高时反接收机的整体性能。由于信道的动态时变,信道当前估计的结果往往无法准确地补偿后续时刻的信道,因此实现对双扩展信道的跟踪是实现可靠有效水声通信的又一关键。针对信道的时变性,本文从自适应滤波理论出发,通过挖掘信道稀疏性和建立动态模型的两种方法来对抗由海洋动态特性带来的信道变化,并进行了仿真验证。前者是将传统自适应算法和稀疏约束问题的代价函数相结合的序贯稀疏估计,本文将已有的零吸引算法引入到扩展函数的跟踪问题中,试图通过稀疏约束降低该跟踪问题的自由度；后者则是直接建立双扩展信道向量的动态演化模型,通过卡尔曼滤波对信道向量进行序贯更新,滤波输出即为每一时刻的双扩展信道向量。最后结合实验数据,论文验证了传统时反均衡接收机用于双扩展信道水声通信的局限性,总结了将本文所述算法融入时反接收机的信号处理结构,并且验证了系统的可行性。实验处理结果表明：在真实环境中,基于互模糊度函数的信道估计算法能以较低的复杂度和较少的训练序列得到优于正交匹配追踪算法的估计结果；为了应对信道的时变引入的动态稀疏序贯跟踪无法收敛,而动态卡尔曼跟踪能够在块处理的基础上解调出信号。更多还原

【Abstract】 Underwater acoustic (UWA) communication technology is the key for solving the transmission and processing of underwater information. In order to improve the system spectrum efficiency and deal with application requirement for high speed underwater acoustic communication, phase modulation technology has replaced the frequency modulation technology. Considering the rapid changes of phase response due to ocean dynamic, phase modulation technology is particularly vulnerable to the impact of the sound propagation environment. The channel impact can be summarized as limited bandwidth, long multipath delay spread, and rapid time variation. The excessive delay spread results in significant inter symbol interference (ISI) while time variation can lead to Doppler effects. The channels which often exhibit both time and frequency dispersion are referred to as doubly spread channels. When there exists large Doppler spread caused by large difference of arrival angles from different propagation paths, Doppler compensation using a single average value is not enough and the joint estima-tion for both delay and Doppler spread is necessary to recover the broadened signal. Moreover, the channel estimation results should be updated in real time to ensure the stability of the receiver when the time variation is severe. Facing the doubly spread and rapidly time-varying channels, this thesis conducts a research incorporating chan-nel estimation and dynamic channel tracking, aiming at realizing reliable acoustic communications.The study on the channel characters is the basis for the analysis and design of communication systems. First of all, the basic KRAKEN model and ray model is ana-lyzed for the way to introduce the Doppler spread. Then, the parameterization channel model is derived from a mathematical model that describe the discrete time channel input-output relationship. Finally, the image method is used to generate discrete mul-tipath according to the locations of transmitted/received sensors and the waveguide environment information. The amplitude of each paths is determined according to propagation attenuation and the autoregressive model which indicates the dynamic character. This dynamic doubly spread channel model is developed for the simulation and analysis of channel estimation and tracking algorithms stated in the thesis.To study the channel character of both delay and Doppler, one can obtain spatial information of environment and location of the transmitter/receiver, and achieve the dynamic evolution of the environment and platform motion information. Therefore, it is effective and complete to study the channel and the channel response to be estimated is the delay-Doppler spread function. Since the estimation of delay-Doppler spread function problem is an underdetermined inverse problem, the traditional estimation approaches do not work. This thesis adopts two approaches to solve the estimation problem. From the perspective of constraint optimization problem, applications of compressive sensing have been extended to delay-Doppler spreading function estima-tion. This thesis chooses the orthogonal matching pursuit (OMP) algorithms to conduct the simulation research. Moreover, an alternative solution is considered from the per-spective of matched filtering. Unlike the sparse channel estimation, the approach is not based on sparseness assumption, and the number of paths is determined simply by thresholding cross-ambiguity function (CAF) based on the noise power. The basic idea is to set rules of selecting the distinct paths on the plane instead of improving the resolution of the CAF. Simulation results validate that the OMP may achieve good estimation precision; however, this greedy algorithm with good estimation precision requires sufficient training symbols and much more computational effort to precisely select the parameters. While the direct CAF-based approach could support determining the number of paths by the noise power adaptively with lower computational complex-ity to achieve more reliable performance.However, the assumption that the channel remains stable over the training time could be easily violated in practice. It means that the channel estimates obtained from training symbols is not suitable for the subsequent information symbols. Hence, chan-nel tracking is the key technology to realize the reliable communication system. Facing the dynamic of channels, this thesis adopts existing sparse adaptive algorithms, which integrates the basic adaptive filter theory and the sparse property of the channel, to track the two dimensional spread function. Moreover, an alternative way is considered that the dynamic model is established to indicate the time-varying of channel and Kal-man filter is used to track the spreading function.Finally, using experimental data collected in shallow water with a mobile trans-mitter, the limitation of the traditional time reversal receiver has proved. The time reversal receiver with the algorithms analyzed in this thesis has summarized. Pro-cessing results show that the CAF-based channel estimation could achieve more robust performance with lower computational complexity and fewer training symbols than OMP method; the sparse sequential tracking algorithm failed in practice, while the dynamic Kalman tracking could demodulate the signal successfully.更多还原