Blind channel identification on CDMA forward link based on dual antenna receiver at handset and cross-relation

This work focuses on the forward link in a CDMA based multi-user communication system. The mobile handset is assumed to have two antennas, either spatially separated or having diverse polarizations. Blind channel identification via the cross-relation method (CRM) of Xu, Liu, Tong, and Kailath (1995) is possible when there is a common input into at least two distinct channels (not having common spectral nulls.) The common input in the forward link is the superposition of all active users: desired user plus multi-user access interference (MAI). Although each user's signal is typically below the noise floor prior to cross-correlating with it's respective code, the power of the sum signal is above the noise floor-well above the noise floor in the case of many active users which is the goal of third generation cellular systems. We present simulations assessing the quality of the channel estimates for two types of equalizers: a zero-forcing equalizer and a RAKE receiver. Initial simulation studies indicate that CR-based blind channel estimation may need to be supplemented by additional a-priori information (such as training sequences or the pilot channel) in order to more closely approach the BER performance obtained when the channels are known exactly.