EFFICIENT IMPLEMENTATION OF LINEAR MULTIUSER DETECTORS FOR ASYNCHRONOUS CDMA SYSTEMS BY LINEAR INTERFERENCE CANCELLATION

The decorrelating and the linear, minimum mean-squared error (MMSE) de tector for asynchronous code-division multiple-access communications a re ideally infinite memory-length detectors. Finite memory approximati ons of these detectors require the inversion of a correlation matrix w hose dimension is given by the product of the number of active users a nd the length of the processing window. With increasing number of acti ve users or increasing length of the processing window, the calculatio n of the inverse may soon become numerically very expensive. In this p aper, we prove that the decorrelating and the linear MMSE detector can both be realized by linear multistage interference cancellation algor ithms with ideally an infinite number of stages. It will be: shown tha t depending on the signal-to-noise ratio, the number of active users, and the choice of the cancellation algorithm, only a few stages are ne cessary to obtain the same BER performance as with the ideal detectors . The computational costs for one stage of a linear interference cance llation algorithm are essentially given by one matrix-vector multiplic ation. Thus, the computational complexity can be reduced considerably. Since each stage introduces a time delay equivalent to the bit durati on, the number of stages also determines the detection delay. Because a few stages are sufficient, this approach can also be used to obtain receiver structures with low memory consumption and detection delay.