Novel approach to multiuser detection for CDMA systems

In this paper a new scheme for multiuser detection is proposed. The detector is based on the Fisher's linear discriminant function whose weights are obtained from periodically transmitted training data. The proposed detector does not require knowledge of the signatures of the remaining users in the system and no assumptions are made in its derivation regarding the noise which corrupts the signal. Also, the same detector can work in both, synchronous and asynchronous systems. The computer simulations show that the detector performs very well in a wide range of signal-to-noise ratios and has excellent performance in cases when the received signals from various users have unequal powers.