Performance evaluation of the independent component analysis method and the analytical constant modulus algorithm for wireless communications

This paper studies the performances for wireless communications of the independent component analysis (ICA), and the analytical constant modulus algorithm (ACMA). These algorithms can be used at the base station to separate the received signals. The ICA algorithm aims to statically whiten the data set for second-order statistics, and for higher-order statistics; while ACMA aims to find the solution in a data set which has a constant modulus. The methods are blind beamformers, so no training sequence is required, and the array of sensors requires neither a good calibration, nor the knowledge of the geometry. The main drawback is the prohibitive computational costs. The methods are used for noisy multipath fading channels with a Rayleigh distribution for the amplitude, and a Laplacian distribution for the direction of arrival. Scenarios are simulated for different signal-to-noise ratio (SNR), and bit error rate (BER) is derived. Lastly, utilization of the methods as beamformers/equalizers is shown.