Complex RPROP-algorithm for neural network equalization of GSM data bursts

摘要: Neural networks have been studied for channel equalization purposes with quite promising results. However, not a lot of published results are available their performance in realistic mobile systems, such as Global System Mobile communications (GSM). In this paper we the use complex-valued multilayer perceptron (MLP) network when transmitting data bursts through GSM-channels and nonlinear channel. addition to conventional complex backpropagation algorithm training network, also presented version Resilient PROPagation (RPROP) algorithm. These methods then compared using GSM models well model. Performance comparisons made terms bit error rates (BERs) computational complexity. Results show that MLP trained RPROP achieves approximately good backpropagation, but clearly smaller load.