Study Of Code-Excited Linear Prediction Based On Genetic Algorithms And A Kind Of Irregular RS Code

Source coding and channel coding are two important aspects of signal processing. With the development of modern digital communication technique, it is an important problem for study of how to transmit more information within limited bandwidth. For the importance of language for people's communication and information changing, compression coding of speech signal naturally become an important research realm for digital processing of speech signal. In the first chapter of this thesis, we first simply introduce the necessity of speech coding and two different ways of coding, and then we have a detailed discussion on the Linear Prediction Coding (LPC) of speech. Linear prediction analysis is a basic technique for digital processing of speech signal and the base for the technique of digital processing of speech signal to be implemented in many applications. It has very high ?value for every realm of speech signal processing. So, in this chapter,the basic principle and algorithm of LPC for speech signal have been discussed in detail. Last, we simply introduce the basic principle of Code-Excited Linear Prediction (CELP) and its special features and advantages on speech signal compression. After speech signal coding has been introduced, we discuss Genetic Algorithms (GAs) in chapter two. In this chapter, we study and discuss the basic thoughts, principles and Simple Genetic Algorithms. After we have comprehended GAs, we have the bases of utilizing GAs to search stimulative code word in CELP. We utilize GAs to search stimulative code word in CELP when we have had the knowledge of speech signal coding and GAs. In chapter three, the simulating results have been presented, the advantages and disadvantages of utilizing GAs in CELP have been analyzed, and the direction for next study has been pointed out.Channel coding (also named error-correcting coding or error-control coding) is another important researching direction for signal processing. The essence of error-correcting coding is to make strong the anti-disturbance ability when signal is being transmitted. From chapter four to five, we simply introduce the basic principles, concepts and methods of error-correcting codes. The concepts and constitution methods of linear block codes, cyclic codes, BCH codes and Reed-Solomon (RS) codes are discussed, as well as their subordinating relationships. For the importance of the relationship between information and parity, the generator matrix and the parity-check matrix of every code have been discussed. Based on the discussion, we constitute a kind of irregular RS code by using the roots of the generator polynomial to define cyclic codes, and using the code as the outer code, together with the inner code selected from one of BCH codes to constitute a kind of concatenated code module. In chapter six, the details of constitution of this concatenated code and the simulatingresults are presented. The feasibility of this kind of concatenated code is supported by the simulating results. For the irregularity and the complexity of this kind of code to be implemented, we need to analyze the capability of this code in theory. It is the aim for next study.