Please use this identifier to cite or link to this item: http://hdl.handle.net/1783.1/566

Blind and group-blind multiuser detectors for reverse link in DS-CDMA systems

Authors Xu, Zhiyu
Issue Date 2002
Summary Direct sequence (DS) code division multiple access (CDMA) has become the standard for digital wireless communications and multiuser detectors (MUD) can substantially increase the DS-CDMA system capacity. Blind MUDs do not need the training sequences and channel coefficients of the users, making it more suitable for applications in bursty traffic environment. Existing blind MUDs assume the receiver knows the information of signature sequence of the desired user only. However, for the reverse link, the receiver at the base station knows the signature sequences of all users within the cell, but not those in neighboring cells. Blind MUDs addressing this scenario are called group-blind MUDs. This thesis studies blind and group-blind MUDs that can effectively suppress not only the intra-cell interference, but also the inter-cell interference that is treated as AWGN noise in conventional receivers. We propose an innovative subspace-based blind noise suppression decorrelating MUD that makes use of the knowledge of the signature sequences of the users within the cell, but without explicitly estimating the channel coefficients. Traditional blind methods perform blind channel estimation and then blind detection using the estimated channel information, making it prone to channel estimation error. It is shown that the proposed blind MUD offers substantial performance gain by bypassing the channel estimation step. We have also applied the interference cancellation technique in multistage group-blind multiuser detectors. We derived a simplified RLS (SRLS) algorithm for a decision directed MMSE detectors and analysed the steady-state SIR performance. Compared with the conventional RLS algorithm, the SRLS algorithm has lower computational complexity, faster convergence speed and higher performance gain. Existing group-blind MUDs generally assume that the number of interfering users from other cells is known, which is unrealistic in practice. We propose a robust adaptive method to estimate the number of strong interferers in color Gaussian noise and show that the group-blind detector performs comparably to the group-blind MUDs with known number of interferers, and even better in low to medium SNR as it only tracks the strong interferers. Higher spectral efficiency can be achieved by using overlapped multicarrier DS-CDMA systems at the expense of inter-carrier interference. We derive the optimal sub-carrier overlap ratio and show that the blind detection technique can suppress the inter-carrier interference effectively.
Note Thesis (Ph.D.)--Hong Kong University of Science and Technology, 2002
Subjects
Language English
Format Thesis
Access
Files in this item:
File Description Size Format
th_redirect.html 337 B HTML