Spatio-temporal array processing for downlink transmission in DS-CDMA systems

We address the problem of optimal downlink transmission in a DS-CDMA system where periodic orthogonal Walsh-Hadamard spread different users’ symbols followed by scrambling by a symbol aperiodic cell specific overlay sequence. We assume that base-stations have multiple-antenna transceivers and each user’s receiver consists of a RAKE. Multiple transmit antennas can reduce the interference and increase the signal-to-noise ratio at mobile receivers. It is assumed that at least partial knowledge of the intracell users’ downlink channels is available at the base-station. This information combined with the knowledge of the type of mobile receiver, suffices to perform optimal spatio-temporal downlink transmission at base-stations. Although spatial and temporal dimensions can no longer be jointly exploited due to the aperiodic overlay scrambling, pure spatial processing or beamforming, can still be performed in order to optimize the whole spatio-temporal processing through the transmitter-channel-receiver cascade. We provide closed-form relations for the signal, interference and noise terms at the output of a mobile station RAKE receiver as functions of the transmit power allocation and beamforming weight vectors. We show that analytical solutions to two optimization problems can be found for the multiuser transmit beamformers, relying on the spatio-temporal structure of the propagation channel. circumstances in [1] it was demonstrated that orthogonality between the spread signals can be restored at each receiver by properly filtering/spreading the symbols intended for different users based upon the information of the channel state associated with each user. Thus, a number of interfering users more than the processing gain may be located in the same cell, in particular accounting for the users in soft hand-off mode. The application of these techniques is not straightforward when the symbol-rate cyclostationarity no longer exists due to the use of aperiodic overlay spreading sequences which spread/randomize the orthogonal user sequences. It has to be noted that, assuming the fading processes slow enough, in the structure of this downlink problem, the only entity fixed over the processing interval is the propagation channel. The actual channel as seen from the base station to a certain user consists of the cascade of spreading, transmit filters, propagation channel, receive filters and RAKE receiver. Due to the aperiodicity of spreading sequences this cascade results in a time-variant filter from symbol to symbol, which precludes the possibility of performing feasible adaptive temporal pre-filtering at the base station, as in [1], because the pre-filters need to be up-dated every symbol period. Some related work is also found in [5].