Performance of Cellular Neural Network Based Channel Equalizers

Abstract—In this paper, a popular dynamic neural network  structure called Cellular Neural Network insert ignore into journalissuearticles values(CNN); is employed  as a channel equalizer in digital communications. It is shown  that, this nonlinear system is capable of suppressing the effect of  intersymbol interference insert ignore into journalissuearticles values(ISI); and the noise at the channel. The  architecture is a small-scaled, simple neural network containing  only 25 neurons insert ignore into journalissuearticles values(cells); with a neighborhood of r = 2 , thus  including only 51 weight coefficients. Furthermore, a special  technique called repetitive codes in equalization process is also  applied to the mentioned CNN based system to show that the  two-dimensional structure of CNN is capable of processing such  signals, where performance improvement is observed. Simulations  are carried out to compare the proposed structures with  minimum mean square error insert ignore into journalissuearticles values(MMSE); and multilayer perceptron  insert ignore into journalissuearticles values(MLP); based equalizers.