A COMPLEXITY MINIMIZATION APPROACH FOR ESTIMATING FADING GAUSSIAN CHANNEL IN CDMA COMMUNICATIONS

R–azvan Cristescu 1 , Jyrki Joutsensalo 2 , Juha Karhunen 1 and Erkki Oja 1
razvan@mail.cis.hut.fi, jyrkij@cs.tut.fi, Juha.Karhunen@hut.fi, Erkki.Oja@hut.fi

A complexity minimization approach has been recently introduced as a generalization of standard Independent Component Analysis. There also time correlations of the source signals are taken into account in addition to the independence assumption. Generally, minimiza­ tion of the complexity measure can be computationally involved, but if the sources are assumed to be Gaussian and time correlated, minimization becomes clearly sim­ pler. In this paper, this simplified method is applied to a timely communications problem, which is estima­ tion of the fading channel coefficients of the desired user in a code­division multiple access (CDMA) system. Simulations with downlink data, propagated through a Rayleigh fading channel, show noticeable performance gains compared with blind minimum mean­square er­ ror channel estimation, which is currently a standard method for solving this problem.