Parallel Interference Cancellation Multiuser Detectors For DS-CDMA Communication Systems

A non-linear sub-optimal multiuser detector in the form of \nparallel interference cancellation (PIC) has been studied. The \nmain objective of this thesis is to develop an analytical model \nfor PIC performance analysis and propose new near-optimum \napproach. \n \nSince the exact performance analysis of PIC is difficult to derive \ndue to its nonlinear decision function, previous work tends to \nadopt computer simulation method or evaluate through Gaussian \napproximation (GA) method. For PIC detector, the GA method may not \napply since there may exist a dominate interference signal. In \naddition, the central limit theorem is not applicable to model the \nresidual MAI in the case of PIC due to its own structural \nproperty. We develop an analytical model to derive the exact BER \nperformance in the case of two users, and extend the method to \napproximate cases when moderate-to-high SINR can be encountered. \n \nWe propose a gradient adaptive parallel interference cancellation \ndetector and investigate its performance. The presented PIC \ndetector is equipped with a set of adaptive weights which are \nadjusted through a new proposed gradient adaptive step size-LMS \n(GASS-LMS) algorithm to reduce the cost of wrong interference \nestimation as existed in the conventional PIC. The initial state \nis deliberately set based on the function of probability of error \nto reflect the reliability of the tentative decision from the \nprevious stage. \n \nWhile the most previous work on MUD are restricted to cases where \nthere is no intersymbol interference (ISI), we consider the \nproblem of joint detection of MAI and ISI, which is crucial to \nenhance the performance of the third and future generation systems \nwith high data rate applications. Simulation results are provided \nto show that our low complexity joint detector can perform very \nwell, yielding the bit error rate (BER) close to the non-ISI \nsingle-user error rate.