Adaptive multi-user detection of fading CDMA channels using antenna arrays

We consider the problem of detecting a number of short-code CDMA signals that impinge (bit-) synchronously on the antenna array of a base station (reverse link). The signals are assumed to be corrupted by both multipath fading and multiple access interference (MAI). To boost performance (measured either in BER for each user or in user capacity), we are interested in using jointly for the signal detection three key signal detection technologies: (1) RAKE processing, (2) array processing, and (3) multi-user detection. The resulting space-time multi-user detectors (ST-MUDs) have been shown in previous works to offer significant improvements over previously existing techniques. In this paper we focus on a specific type of linear detector presented by Huang and Papadias (see Proc. 1998 PIMRC Conference, Boston, 1998), which has been shown to combine good performance with implementation advantages. We propose to update this linear detector with adaptive filtering algorithms, and in particular with stochastic-gradient type trained or blind algorithms. It is shown that, such simple techniques allow the receiver to approach the ST-MUD MMSE solution and achieve significant capacity gains with respect to conventional techniques.