Fast Gaussian Process Regression for Multiuser Detection in DS-CDMA

Recently, Gaussian process has been excellent for proving the effectiveness in solving the multiuser detection issue in code division multiple access system. Even with limited training sequences, Gaussian process-based solutions still surpass other approaches. However, due to the high complexity in terms of computation, the performance of this approach might be degraded. In this letter, we would like to propose an efficient method to reduce the complexity while still maintaining the desired accuracy. Finally, the proposed method is validated via the experiment.