Research On Channel Blind Equalization Based On Dynamic Fuzzy Neural Network

In the modern digital communication system, as the channel's non-ideal characteristic caused by the ISI is a major factor which affects the quality of communication. In order to overcome the ISI, needing to join the equalizer on the receiving end, to compensate for the characteristics of the channel, and to resume the transmit sequence correctly. Traditional adaptive equalizer need to send the training sequence, which will affect the efficiency of communications. Blind equalization as a emerging equalization technology, be able to need not to use the training sequence, just using the statistical characters of received signal sequence itself to equalize the channel. At present, it has been widely used in communications, radar, seismic and image processing and other fields. As a result, the study has important theoretical significance and use value.Firstly, this paper describes the basic principles and development of the blind equalization technology, introduces the basic blind equalization algorithm, analysis its strengths and weaknesses by simulation.Secondly, expounds the theory and characteristics of fuzzy theory and neural network systematically, compares the operational mechanism of theirs, analyses the inevitability of the combination of them, and then introduces the structures of the fuzzy neural network, focuses on the analysing of dynamic fuzzy neural network. Through the analysis of selecting the network's activation function and membership functions of network as well as the network structure, summarizing the Characters of the dynamic fuzzy neural network, and validates the good performance of dynamic fuzzy neural networks in nonlinear function approximation by simulating comparation, provides the theory evidence for dynamic fuzzy neural network used in the channel blind equalization.Finally, applicates the dynamic fuzzy neural network in the channel blind equalization, uses the good structure and the capacity concerning non-linear approximation of dynamic fuzzy neural network, by changing the cost function,designing the blind equalization algorithm based on dynamic fuzzy neural network, and list the to the algorithm formula, by computer simulation shows the effectiveness of the algorithm.