Detecting active channels in an asynchronous DS-CDMA multiuser receiver

Abstract

Nearly all the work in multiuser detection for direct sequence code division multiple access (DS-DCMA) assumes that the active channels are known. Previous attempts to solve the active channel detection problem assume short codes and use subspace methods that treat the multiple access interference (MAI) as noise. We consider forward link IS-95 asynchronous long code multiuser receivers and design an active channel detector that estimates the MAI as an intermediate step. The resulting algorithm subsumes multiuser detection in that we estimate symbols, amplitudes, and active channels jointly. The detector iteratively estimates and removes the MAI using an approach combining order statistics and constant false alarm rate thresholds on mixture densities. The performance of the new detector is compared to a conventional active channel detector which uses only the matched filter sufficient statistic. The new detector provides much higher probability of detection as a function of both capacity (number of active users) and SNR for cases of uniform and geometric power distributions. The new detector is shown to achieve nearly perfect probability of detection even in the near-far case with a moderately loaded system and just a few symbols.