基于概率图模型的水声通信信道建模及迭代信号处理

项目名称： 基于概率图模型的水声通信信道建模及迭代信号处理

项目编号： No.61471351

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 无线电电子学、电信技术

项目作者： 武岩波

作者单位： 中国科学院声学研究所

项目金额： 83万元

中文摘要： 水声通信信道是复杂的、多变的信道，影响信道的因素有很多，不同信道状态下有着与之对应的典型信号处理方式和处理参数。本申请拟基于概率图模型研究水声信道信号特征与物理因素之间的概率约束关系，并研究基于此模型研究水声通信系统整体的迭代信号处理，进而使通信信号处理适用于不同的信道状态。信道图模型的约束可以认为是声速梯度特征、载体的位置、载体运动速度等物理参量与信道的时变特性、多径大小、空间相关性、多普勒变化率等信号表象之间的关系。 研究内容包括，根据水声信道的物理规律建立广泛适用的水声通信信道的概率图模型，并结合实测数据进行模型的改进和学习，结合信道传输函数和信道冗余编码的概率图模型，优化整体的模型描述，实现整体图模型的迭代处理。与现有的迭代均衡方式相比，可更好的利用水声信道的约束特征，如果信道的物理参量是先验信息可以直接用在通信处理中，如果没有这些先验信息可以通过迭代处理逐渐恢复出其概率描述。

中文关键词： 水声通信；水声信道；信道均衡；概率图模型；和积算法

英文摘要： Underwater acoustic communication channel is complex and time-varying, and there are many physical factors effect on it. The classic signal processing method and its parameters for underwater acoustic communication need to be properly chosen depending on the channel states. The study of factors effecting on the underwater acoustic channel based on probabilistic graphical models is suggested in this research proposal, and the research area is further expanded to the iterative signal processing for underwater acoustic communication based on the graphic channel model. The channel physical parameters include sound speed profile characteristic, the relative position and velocity of the communication devices. By using the channel constrain, the channel effects on the communication signal, such as multipath, time-varying, spatial correlation, and Doppler phenomenon are derived from the channel physical parameters. The research contents are introduced as follows. The probabilistic graphical model of underwater acoustic communication channel is built based on the physical laws and constrains of underwater acoustic channel, and the model is further improved by machinery learning from the real underwater acoustic signal samples. The channel graph model and the probabilistic graphical model of channel code are combined and optimized to describe the whole communication system, and the iterative processing based on the whole graphical model is derived. Compared with existing iterative equalization method, the proposed method has the benefit of utilizing the channel constraint characteristics. The prior information of channel physical parameters can be directly used in the communication signal processing, and if no prior information, the distributions of channel physical parameters are gradually recovered by iterative processing.

英文关键词： underwater acoustic communication;underwater acoustic channel;channel equalization;probabilistic graphical model;sum-product algorithm