Channel estimation and multiuser detection for long code CDMA

Abstract

In a long code CDMA s)rrrem subspace technique is directly applied to downlink channel estimation when spa-riaUtempora1 diversiry is available. Detection of the desired user is pe$ormed by Vpiral linear techniques after channel is estimated. Since the method depends on eigenvalue decomposition on the data covariance matrix which is typically estimated fmm fnite data samples, the optimal solution is then perturbed. Therefore, statistical performance of both the channel estimator and detectors is analyzed based on apenurbarion technique. It isfurrhefjusr$ed by numerical examples.