基于香农映射定理的语音信号信源信道联合编码的分析-analysis of joint source-channel coding of voice signals based on shannon mapping theorem.docx

摘要随着通信技术的迅猛发展，频谱资源日益紧缺，节约系统占用频谱的问题越来 越被科学家们所重视。本课题采用近似模拟信源信道联合编码技术结合正交频分 复用技术，使系统所占带宽得到了显著的降低。本文在了解了语音编码技术历史、发展以及当前技术和背景的基础上，根据香 农编码理论和信源信道联合编码原理，在 MatLab 环境下对 PCM 编码和信源信道 联合编码技术进行了仿真；研究了不同空间映射曲线在联合编码中的性能及其可 行性，优化了信源信道联合编码。课题首先比较两种编码技术的性能，一种是传统的数字 PCM 编码调制技术； 另一种是本文所提出的近似无量化的直接采样方法——线性映射信源信道联合编 码器。近似无量化直接采样是把采样信号近似无量化信号直接映射到一个星座图 上，这样不但简单而且能减少占用的带宽，所以该技术也是一种近似模拟的方法。 通过详细比较二者性能，可以得到结论，这种简化的联合编码系统不仅带宽占用 小，而且性能更为优良。另一项主要工作是筛选合适的空间映射曲线，使语音采 样信号能够直接映射到信道中去。文中除了使用客观参数比较性能，我们还引入 了主观评价的方法。通过分析采用这些映射曲线的编码器性能，能够找到更加适 合在语音编码中应用的曲线。通过比较发现，本文所采用的线性映射方法在占用单倍带宽的情况下，可以得 到更好的性能，因此可以实现降低带宽的目标。另外，采用两种空间映射曲线将 信源样本点直接映射到信道空间简化了系统，降低了带宽。关键词：香农定理，语音信号，信源信道联合编码，映射ABSTRACTAs the communication technologies developing significantly, the available frequency spectrum are more and more lacking. Reducing the occupied spectrum of the system becomes very important. This project applies near analog joint source-channel coding technology and combines OFDM method in order to reduce the bandwidth significantly.Based on realizing the history, development and current statement of coding technology, this project simulates the PCM coding and joint source-channel coding technologies developed from Shannon coding theorem under MatLab environment, and also compares the advantages and disadvantages. Research different space mapping curves and the feasibility, and optimize the joint source-channel coding.One of the works in this project is to compare two techniques. One is the traditionaldigital tandem source channel coding; the other is a near analog joint source-channel coding method. For the digital tandem coding method, it applies the ??-law PCMtechnique as the source encoder and digital modulation as the channel encoder, and these two encoders are independent to each other. The near analog joint source-channel coding method operates on the samples directly without quantization. It just maps the samples into a linear constellation, which is a simple but less bandwidth occupied technique. Another work in th