Research On The Neural Based Blind Equalizer In Digital Communication

In modern communication system, the Inter-symbol Interference (ISI) caused bynon-ideal character of channel is the main factor which affects communication quality.As an effective method to restrain the ISI equalization technology is becoming theresearch hotspot in the world. Compare to the classical equalizer，blind equalizerwithout training sequence is more benefit to modern communication system. And ithas been applied to many areas, such as communication, radar, earthquakeprospecting, and image processing.In this paper, the development status quo of the blind equalization technology athome and abroad is analyzed systematically. The function of the equalization andblind equalization algorithm theory are introduced. Then a proved dual-modealgorithm is proposed based on analyzing the advantages and drawbacks of existingtraditional algorithms. This algorithm is a kind of automatic switching algorithm withan algorithm reset module. It adopts MMA algorithm in the beginning, and adoptsMDD algorithm when the algorithm is converged. Well, several simulations are madeto test the equalizer’s performance.The principle and structure of the feed-forward neural network are deducedgradually, and the algorithm iterative process and neural network equalizers baseddifferent cost functions are discussed in this paper. Then a new equalizer is designedby applying novel algorithm to neural network structure, and simulation results haveproved the effectiveness of the proposed algorithm, the convergence of high orderQAM signal and the tracing capability of mutant channel are tested. Finally, thefurther research direction was pointed out, which may be of some assistance tosubsequent research work.