Blind Equalization Based on Wavelet Neural Network Optimizing by Genetic Algorithm

Blind equalization based on wavelet neural network optimizing by genetic algorithm was proposed for the conventional gradient algorithm is sensitive to the values of the initial parameters. At beginning, a segment finite data was collected for genetic algorithm to get a group of asymptotically optimal initial parameters. And then, gradient-descent algorithm was adopted to train network to trace and compensate the channel characteristic to implement equalization. Convergence and stability analysis of the proposed algorithm is also provided. The goodness of the proposed blind equalization algorithm is demonstrated with the aid of a simulated the non-linear channel.