Nonlinear receiver concepts in reduced dimensions for WCDMA

Nonlinear receivers for WCDMA, such as maximum likelihood schemes, promise optimum performance at the expense of high computational complexity. We combine linear and nonlinear receivers to obtain a good trade-off between low complexity and optimum performance. Unlike other schemes for dimension reduction the proposed architecture distinguishes between long-term and short-term effects in the wireless communication channel. Our adaptive linear prefiltering stage reduces channel dimensions based on accurate estimates of the slowly changing spatial and temporal channel characteristics (long-term properties). In the second stage any receiver concept for antenna arrays and DS-CDMA systems can be used, where we employ a maximum likelihood (ML) channel estimator and detector. Due to dimension reduction the estimation quality of the fast fading complex amplitudes (short-term properties) in the 2nd stage is improved and the complexity of the ML receiver is reduced by 50-75% with small impact on performance in typical uplink scenarios. We also discuss and illustrate the consequences of dimension reduction on short-term processing, Finally, simulations in different scenarios for the WCDMA uplink demonstrate the complexity gain, improved channel estimation, and small performance degradation.