Blind identification and equalization based on maximum kurtosis criteria and genetic algorithm

Based on maximum kurtosis criteria, a new blind identification and equalization algorithm is designed for linear system. When system parameters are evaluated, the equalizer parameters can also be obtained at the same time by searching inverse filter parameters. The evaluated values of system parameters are constantly regulated by maximum kurtosis criteria so as to approach the real values. Because of the utilization of high order cumulant, this algorithm can effectively suppress Gaussian noise. The real coded genetic algorithm is also proposed to search the optimum solution, which can overcome the drawback of traditional gradient search technique which is likely to fall in local minimum. Simulation results demonstrate that the algorithm not only has a fast convergence performance and high accuracy, but also can improve the output SNR greatly.