It has been shown that the performance of linear detectors for DS-CDMA varies
widely depending on certain system parameters. In general
it would seem that an increase in complexity corresponds to an
increase in the level of performance in the worse case scenarios of
channel ISI and asynchronism. The
least complex of the detectors studied, the matched filter, performs
poorly under heavy channel ISI, asynchronism, non-orthogonal spreading
codes, or any combination of the three. The zero-forcing detector
performs very near the MMSE detector, however, there becomes an
implementation problem when 
does not exist. The MMSE detector performs very
well under asynchronism and channel ISI, but requires knowledge of the
channel and user timing which may not be available.
The adaptive algorithms have been proposed as an alternative to the fixed detector implementations each at a certain cost. The LMS algorithm converges to the MMSE solution, but the choice of stepsize, training bandwidth, and user power all effect the convergence properties of the algorithm. In a time varying, mobile system, the LMS algorithm under ideal parameter selection may still be too slow to be implemented. Also shown was the performance of a blind adaptive implementation of linear detection using the Constant Modulus Algorithm. This method requires only knowledge of the user spreading codes to be effective and is resistant to the near-far effect and asynchronism, but requires an estimate of a fourth order statistic.
The software tools' versatility allows a variety of simulations to be run, only a few of which have been presented in this report. The software has been designed with numerous user specified parameters for maximum utility. The Matlab m-files can also be used to run Monte Carlo simulations if so desired. Not all possible linear detectors are addressed in this project, and for this reason the software was created to be modular and upgradeable. The purpose of the software is to provide a platform for testing new DS-CDMA detectors against the established fixed designs, namely the matched filter, zero-forcing detector and MMSE detector. The simulation results shown demonstrate the performance of the software as a DS-CDMA linear detector evaluation platform. In the future, it is possible that non-linear detectors and other adaptive algorithms will be implemented in the software as well.