Blind Equalization Algorithms Based On Wavelet Transform And Support Vector Machine

In the wireless communication system, due to complex characteristics such as multipath fading of channel, the inevitable inter-symbol interference in the signal transmission process is always generated, and the quality of communication was seriously affected. The emergence of equalization technology can solve this problem, it make the communication system can be more reliable and effective. The blind equalization technique became a hotspot in recent years due to its many advantages in performance. Based on the analysis of the various blind equalization algorithm, aim to small samples for communications, with support vector machines for mathematical tool, combining with wavelet transform, fractionally-spaced, as well as the decision directed algorithm, blind equalization algorithm was studied in this thesis. the following algorithm was proposed and all of the algorithm was simulated through matlab, the performance of various algorithms was also analyze, the content of this paper are as follows:1. support vector machine blind equalizer based on orthogonal wavelet transformationSupport vector machine is a good kind of learning tool for small samples, it can be used in regression problem or classification, due to its global optimization features, support vector machine algorithm can avoid local minimum. In fact, Blind equalization algorithm is a regression problem on the received signal based on the output signals meeting the error performance. Aim to the shortcomings of traditional constant modulus algorithm, this paper have do some research on SVM blind equalization algorithm based on the characteristics of SVM. After the introduction of wavelet transform to the blind equalization algorithm, support vector machine blind equalizer based on orthogonal wavelet transformation was proposed. The method used the wavelet transformation to enhance the convergence performance of SVM blind equalization algorithm, it have make full use of the global random search feature of SVM. Simulation results shows the superiority of the proposed algorithm2 A blind equalization algorithm based on fractionally spaced orthogonal wavelet transform with support vector machineSince SVM blind equalization search for the optimize weight vector through the linear programming, and the search method show a good performance in dealing with small sample data. However, when the sample data continuously increases, the amount of computation will be greatly increased. The search method of blind equalization based on fractionally spaced orthogonal wavelet transform is stochastic gradient descent method, this algorithm is prone to local minimum, so it is necessary to propose a algorithm to avoid local minimum. According to the above advantages and disadvantages of the two algorithms, A blind equalization algorithm based on fractionally spaced orthogonal wavelet transform with support vector machine was proposed, this algorithms utilizes a short initial data segment to initialize weight vector of blind equalization based on fractionally spaced orthogonal wavelet transform., the algorithm avoids the local minimum and can achieve real-time equalize. Simulation results show that the algorithm has excellent performance3 A Blind Equalization Based on Decision Directed Combined Orthogonal Wavelet Transform Algorithm with support vector machineTo further improve the algorithm's steady-state error performance, a soft-switching dual-mode blind equalization algorithm is proposed, this algorithm is a combination of constant modulus algorithm (CMA) and decision directed (Decision Directed, DD) algorithm. The convergence rate of Liner Equalizer is fast, but the stead error is too high,and it can not correct the phase rotation. Due to the initialization of the weight vector is still more sensitive, its weight vector settings requires further study.Through analysis, A Blind Equalization Based on Decision Directed Combined Orthogonal Wavelet Transform Algorithm with support vector machine is proposed aiming to the two algorithm.Underwater acoustic channel simulation results has proven the good performance of this algorithm.