The Research Of Blind Equalization Algorithms Based On Fuzzy Control

Blind equalization is an adaptive equalization technique, which can equalize the properties of the channel just using the statistic properties of the received signal and overcome the disadvantages of conventional adaptive equalization techniques which require a training sequence and reduce effective information rate of system transmission. Thus, it has become one of the major focus in the field of digital communication and signal and information progress.In blind equalization algorithms, the choice of iterative step-size is very important. If the iterative step-size is too small, the iterative calculation speed is too slow, the stable residual error is small after the algorithm converges . If it is too big, then there will have the vibration near the convergence point although the convergence rate is accelerated. Therefore, the convergence rate and convergence precision restrict each other. Momentum blind equalization algorithm can effectively prevent the algorithm from falling into local minimum, accelerate the convergence process of blind equalization in the process of updating the weights. But in the algorithm, fixed momentum blind equalization algorithm can not adjust the momentum size in time, when the algorithm converges. So it will have greater stable residual error, resulting in bit error rate increasing. As a branch of intelligent control, fuzzy control has been widely applied in many real control problems in recent years. Especially it is often used to control some systems which are non-linear or uncertain or even can not be made to build exact mathematic model effectively and accurately. This paper introduces the fuzzy controller to control iterative step-size and momentum of blind equalization algorithm, it has important theoretical significance and the practical value to enhance the accuracy of convergence and speeds up the convergence rate.The major works of this paper are summarized as follows:(1) This paper briefly summarizes each realization method and its development tendency in blind equalization algorithms, analyzes their characters and deficiency. At the same time, it introduces fuzzy control principles and several kinds of feasible applying ways of blind equalization algorithm.(2) In view of CMA when chooses the bigger iterative step-size, convergence rate quicker, the mean square error is bigger, When uses the smaller iterative step-size, convergence precision is higher, but at the cost of sacrificing the convergence rate. In order to solve the contradiction between convergence rate and convergence precision, this paper proposes three controllers based on fuzzy to control step-size factor of blind equalization respectively. In the algorithm, uses experience knowledge extract information from MSE to control the step-size of CMA, induces algorithm iterative formulas. The simulation results show that CMA has better convergence performance after comparison between these three algorithms and conventional CMA algorithm.(3) It analyzes the momentum's effects on the performance of algorithm, which is in blind equalization algorithm based on momentum forward neural network. It promotes convergence irate effectively, reduces steady residual error and bit error rate through changing momentum in algorithm iteration. Thus, this paper proposes three kinds of fuzzy controllers include fuzzy controller, fuzzy adaptive controller and adaptive neural fuzzy Inference system to control momentum of blind equalization algorithm respectively. The simulation results show that these algorithms' convergence rate speeds up, stable residual error and bit error rate slows down after comparison between these three algorithms and blind equalization algorithm based on momentum forward neural network.