Soft Computing Methods And Its Applied Research In Communication And Information Processing

With the development of communication, computer and electronics, the engineering applications have become more widely and more deeply, the number of mathematics methods increases rapidly in abundant engineering applications. Because of the need of information processing, soft computing methods are used widely. The theory of soft computing developed rapidly, and some other relative theories appear constantly, which increases its importance in the application of information processing. This thesis investigates the use of soft computing methods (including: wavelet theory, artificial neural network, optimization algorithms, fractal theory, chaos theory and mathematical morphology, etc) in information processing, reveals the problems in engineering applications of soft computing methods and combines different mathematics methods in information soft computing processing.The communication channel is the key part of researching communication system, its model and real capacity performance are important research hotspots. I investigate the multi-path time-varying channel model, compare the three diversity techniques: multi-antenna space diversity, channel coding time diversity and multi-carrier modulation frequency diversity, and put forward relative research hotspots and their developing trend.Due to the characters of multi-path time-varying channel and the influence of carrier frequency offset, the channel estimation has become necessary in communication system. With the appearing of non-blind, semi-blind and blind channel tracking methods, especially the abundant applications of multi-antenna space diversity increase the complexity of communication system and improve the system performance at the same time. I use adaptive filter method to estimate the multi-path time-varying channel. Furthermore, I investigate the non-blind channel estimation algorithm by using Kalman filter in multi-antenna channel, and propose a Kalman tracking estimation algorithm in multi-antenna fast varying channel which can be used for image communication transmission system to track the channel effectively. After studying Monte Carlo algorithm, I propose a new multi-antenna time-varying channel blind estimation algorithm, which can track the fast varying channel efficiently without pilot and in the system frequency offset scenarios.Wavelet theory and its engineering applications develop rapidly in recent years, In this thesis, I analyze the theory of wavelet, multi-band wavelet and complex wavelet, and apply them into communication system, signal processing and image processing respectively. Thesis presents an OFDM channel estimation algorithm based on wavelet denoising and wavelet interpolation, which can increase the accuracy of channel estimation and decrease the bit error ratio of OFDM system significantly. Thesis proposes a wavelet scale-varying threshold Wiener filter algorithm to improve denoising performance and adopt wavelet theory to estimate the power spectrum of speech signal effectively. Thesis applies multi-band wavelet and complex wavelet into the embedding of image watermark to enhance the information security and copyright, applies multi-band wavelet and complex wavelet for direct sequence spread spectrumcommunication system and put forward two narrow band interference suppression algorithms. Thesis adopts wavelet and fractal theories in image interpolation enlarging to enhance the accuracy in image enlarging process.To get the optimal solution is a usual case in engineering problems. Optimization theory has become an important part of computing intelligence research, and its application spreads in all research areas of information processing. This thesis adopts the genetic algorithm and particle swarm optimization algorithm into the optimal threshold estimation algorithm of wavelet denoising, and applies artificial fish swarm algorithm for the determination of wavelet threshold and FIR filter design.Multi-carrier modulation can enhance the data rate of system transmission significantly and the OFDM technique has been a research hotspot. This thesis adopts orthogonal wavelet base into multi-carrier modulation, proposes a large carrier number wavelet base modulation algorithm based on wavelet packet, and compares with the OFDM system in Rayleigh complex varying channel. I analyze the technique advantages of wavelet multi-carrier modulation, and its theory and application have become research hotspots.The multi-path effect of communication channel would result in inter-symbol interference (ISI) in communication system, while channel equalization technique can suppress ISI efficiently. In this thesis, the wavelet neural network is investigated, channel equalization is adopted and a scale-varying wavelet neural network channel equalization algorithm is proposed, which can reduce the neural cell number of the network significantly, enhance the equalization quality and decrease the bit error ratio of the system.The communication channel coding is the research hotspot in communication area. In this thesis, I combine the current TURBO code and the chaos theory, and propose the chaos TURBO code/decode scheme. Considering the important effect of interleave in TURBO coding, I investigate and compare the performances of chaos interleave in TURBO coding and the usual random interleave.After studying the theory of mathematical morphology and its applications in image processing, I analyze the performance of image edge detection after using the opening and closing operation of mathematical morphology. An image edge detection algorithm is proposed based on multi-grade mean of mathematical morphology, which can enhance the detecting performance of image edge, and improve the denoising ability significantly.