Multiwavelet Fuzzy Neural Networks Blind Equalization Algorithm

At present, the report on the research of channel blind equalization processing, which combined the Multi-wavelet transform with blind equalization technology, is very few at home and abroad. But it is a very significant topic to research on channel blind equalization technology by means of Multi-wavelet theory, Neural network, Fuzzy theory and making full use of their superiority, and it is also an advanced topic in modern signal processing field. Aiming at the disadvantage of CMA (Constant Modulus Algorithm), the blind equalization algorithm based on Multi-wavelet Fuzzy neural network are studied deep in this dissertation. The main contributions are as follows:1. A balance Orthogonal Multi-Wavelet Transform based on Constant Modulus Blind Equalization Algorithm is proposed on the basis of the structure of blind equalization based on Balance Orthogonal Multi-wavelet Transform and the Orthogonal Multi-wavelet Transform Matrices are achieved. The proposed algorithm takes advantage of multi-wavelet to speed up the convergence rate and the equilibration of Multi-wavelet overcome its disadvantage of prefilter when used. Thus the property of Multi-wavelet is reservation. To improve the performance of algorithm, a Dual-Mode blind equalization Algorithm based on Balanced Orthogonal Multi-Wavelet Transform is proposed by introduce the Dual-Mode algorithm, which is exchange automatically between two Dual-Modes, and the residual error is reduced. The efficiency of the algorithm is proved by computer simulations.2. A Fuzzy Neural Network Blind Equalization algorithm based on Radial Basis Function (RBF) is proposed by using RBF as blind equalizer and introducing fuzzy c-means cluster algorithm. The proposed algorithm makes use of the advantage of RBF and fuzzy c-means cluster algorithm, a good performance is achieved. To QAM signals, a signal transformation ways is research by analysis the traditional methods which divide the real and imaginary part and a fuzzy neural network blind equalization algorithm based on signal transformation is proposed. The simulation results show that a relatively low mean square error and fast convergence rate has been achieved.3. In order to overcome the contradiction between the convergence rate and accuracy for using fixed steps in traditional Constant Modulus Algorithm (CMA), a Fuzzy Neural Networks Blind Equalization Algorithm based on balance Orthogonal Multi-Wavelet Transform (MWT-FNN-BEA) is proposed. In the proposed algorithm, the contradiction between the convergence rate and accuracy is solved by making use of fuzzy neural networks controller to adjust the step of algorithm automatic. Furthermore, by taking balance orthogonal multi-wavelet transform to the input signals of equalizer, the performance of algorithm is further improved. Theoretic analysis and computer simulations demonstrate that the presented algorithm has fast convergence rates, small steady-state error and good noise immunity.Figure [43] Table [6] Reference [79]...