Independent Component Analysis Applications in Cdma Systems

Abstract

Blind source separation (BSS) methods, independent component analysis (ICA) and independent factor analysis (IFA) are used for detecting the signal coming to a mobile user which is subject to multiple access interference in a CDMA downlink communication. When CDMA models are studied for different channel characteristics, it is seen that they are similar with BSS/ICA models. It is also showed that if ICA is applied to these CDMA models, desired user.s signal can be estimated successfully without channel information and other users. code sequences. ICA detector is compared with matched filter detector and other conventional detectors using simulation results and it is seen that ICA has some advantages over the other methods.The other BSS method, IFA is applied to basic CDMA downlink model. Since IFA has some convergence and speed problems when the number of sources is large, firstly basic CDMA model with ideal channel assumption is used in IFA application.With simulation of ideal CDMA channel, IFA is compared with ICA and matched filter.Furthermore, Pearson System-based ICA (PS-ICA) method is used forestimating non-Gaussian multipath fading channel coefficients. Considering some fading channel measurements showing that the fading channel coefficients may have an impulsive nature, these coefficients are modeled with an -stable distribution whose shape parameter takes values close to 2 which makes the distributions slightly impulsive. Simulation results are obtained to compare PS-ICA with classical ICA.Also IFA is applied to the single path CDMA downlink model to estimate fading channel by using the advantage of IFA which is the capability to estimate sources with wide class of distributions.