Equalization for a Wireless ATM Channel with a Recurrent Neural Network Pruned by Genetic Algorithm

A new method for pruning the complex bilinear recurrent neural network(CBLRNN) is proposed in this paper. The pruned CBLRNN is applied to the equalization of signals for a wireless ATM network.The transmitted signal is assumed to be modulated by phase shift keying approach. The pruned CBLRNN based equalizer is compared with currently used decision feedback equalizer (DFE), Volterra filter based equalizer, and multilayer perceptron neural network equalizer. Experiments show that the pruned CBLRNN gives better results in terms of MSE and SER criteria over conventional equalizers.