Adaptive interference suppression for direct sequence CDMA over severely time-varying channels

Recent work has shown that minimum mean squared error (MMSE) based adaptive interference suppression schemes have the potential of providing large performance and capacity gains over conventional reception for direct sequence (DS) CDMA systems. However, the applicability of such schemes to future cellular and PCS systems will depend on their ability to cope with the severe time variations characteristic of a wireless channel. We address this issue in the context of Rayleigh faded DS-CDMA signals. Our main results are as follows. (a) Standard decision directed (DD) MMSE adaptation following some initial training is unable to track a typical fading channel. (b) A modified DD (MDD) MMSE adaptation that exploits differential encoding and decoding to simplify the task of tracking is robust against fading and the near-far problem. (c) MDD-MMSE adaptation survives the arrival of new interfering transmissions even in the presence of a severe near-far problem. (d) The performance of the MDD-MMSE algorithm shows some deterioration when there are several rapidly fading independent multipath components, so that further investigation is needed for devising adaptive algorithms for this setting.