Joint Source-Channel Coding For Speech Signal Based On Shannon Mapping Theorem

As 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 traditional digital tandem source channel coding; the other is a near analog joint source-channel coding method. For the digital tandem coding method, it applies theμ-law PCM technique 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 this project is to develop a proper space-filling curve to map the speech signal samples into the channel space directly. It is another near analog joint source-channel coding technique. According to the analysis of the mapping using these curves, it can be concluded whether it is suitable for speech coding application or not and which one is better. Otherwise, in order to get a reasonable result, not only the objective measurement is used in the project, but also the subjective evaluation is applied.According to the comparison, it can be found that the linear mapping method applied in this thesis has better performance when there is only one time bandwidth, so it indeed reduces the occupied bandwidth and simplified the system. Moreover, the other two joint encoders using two space mapping curves map the speech samples into the channel space directly, which also reduce the bandwidth and simplify the system.