Design of Supervised and Blind Channel Equalizer Based on Moth-Flame Optimization

Abstract

With the increase in demand of high-speed data, the communication channels are suffering from inter-symbol interference, nonlinearity and noise. In this paper, digital channel equalizers are developed for finite impulse response (FIR) and infinite impulse response (IIR) channels. Supervised learning based on minimum mean square error is opted for FIR channels affected by noise and nonlinearity. For IIR channel, blind equalization based on constant modulus technique is employed. The weights of both type of equalizers are determined by a recently developed nature-inspired algorithm moth-flame optimization (MFO). The performance is compared with the same structure of equalizer trained with standard algorithms like: artificial immune system, adaptive particle swarm optimization, binary genetic algorithm and least mean square algorithm. Simulation studies are demonstrated for equalizers designed to compensate the response of two FIR channels, and both are affected by two nonlinearities with noise and three IIR channels. Superior performance by the proposed MFO equalizer is reported based on frequency response analysis, mean square error during training and BER plots during testing.