Performance of adaptive interference suppression for DS-CDMA with a time-varying user population

Abstract

The performance of adaptive interference suppression algorithms for DS-CDMA is studied with a stochastically varying user population. A multiple cell system is assumed with stochastic arrivals and departures of asynchronous users. The adaptive algorithms considered are based on the minimum mean squared error criterion. A tapped-delay line filter implementation is assumed, where the filter spans a single symbol interval. Adaptive algorithms simulated are the stochastic gradient (LMS), and recursive and block least squares. Both decision-directed and recently proposed blind algorithms, which do not require a training sequence, are compared. In addition, we present a variation on the block least squares algorithm in which the received vectors are first projected onto components of the signal space (eigenvectors) with the largest energy (eigenvalues). The purpose of the projection is to eliminate low-level background interference and noise. Computer simulation is used to obtain error rates and time-averaged signal-to-interference ratio (SIR) as a function of traffic load.