MULTIUSER DETECTION IN DS/CDMA COMMUNICATIONS

Abstract

Two novel multiuser detector schemes are proposed and analysed in this thesis. The first scheme is the error control decision feedback (ECDF) multiuser detector, which possesses the immunity against error propagation. The ECDF multiuser detector implements multistage detection by employing an error control successive interference cancellation detector (EC-SIC) as the first stage. The EC-SIC detector is very different from the conventional SIC multiuser detector in that its successive cancellation loop can be cut off to prevent error propagation whenever any decision error occurs. In addition to the immunity against error propagation, the ECDF multiuser detector retains the advantages of the SIC multiuser detector and the multistage multiuser detector. The performance of the ECDF multiuser detector is studied by computer simulations for synchronous DS/CDMA systems. The results show that the ECDF multiuser detector exhibits a significant improvement over many reported schemes. Although the ECDF multiuser detector is a bit more complex than some reported schemes, its complexity is still linear to the number of users. The second scheme is the adaptive blocked MMSE (B-MMSE) multiuser detector for asynchronous OS/CDMA systems. In the B-MMSE multiuser detector, with silence bits inserted, the received signal can be divided into blocks without introducing any mutual interference among the blocks. It is shown that the MMSE detection can be performed optimally within any block independently. Since the block length in the BMMSE multiuser detector is controllable, it provides great flexibility to implement in practice. In order to track the time-varying channel and avoid estimating the users' powers, the B-MMSE multiuser detector is implemented in an adaptive way. The adaptive B-MMSE detector consists of a bank of adaptive Wiener filters which can work together independently. The BER performance and convergence properties of the adaptive B-MMSE multiuser detector have been analysed. Some numerical results in the steady state are presented and compared with other linear multiuser detectors. It is shown that the adaptive B-MMSE detector retains the advantages of the MMSE detection and gives excellent performance. And also extensive simulations are performed to investigate the transient properties of the adaptive B-MMSE multiuser detector. Some suggestions on choosing system parameters are given. The hardware complexity of the adaptive B-MMSE multiuser detector is linear to the user number and block length, while the computational complexity is proportional to the squares of the user number and block length.