Design Linear Multiuser Transmitters from Linear Multiuser Receivers

Abstract

Novel concepts are introduced for finding the relationship between multiuser detection (MUD) and multiuser transmission (MUT), so that the study in MUT can benefit from the well-built theory in MUD. Our study shows that, for any given linear MUD scheme, there exists a counterpart linear MUT scheme, which can be readily designed from the original linear MUD. Based on our observations, a range of MUT algorithms are derived in correspondence with the well-known linear processing schemes in the fields of MUD and array processing, including minimum mean-square error (MMSE), minimum variance distortionless response (MVDR), minimum power distortionless response (MPDR) and maximum signal-to-interference-plus-noise ratio (MSINR), respectively. Furthermore, the physical principles behind the considered MUT schemes are explained with the aid of the concepts built in this contribution. In order to gain insight into the effect of noise power on the achievable performance of a MUT scheme, a noise-suppression factor is introduced to those MUT schemes, such as MMSE, requiring the knowledge of noise variance. Our study shows that the achievable performance of the MUT is generally loosely depended on the noise-suppression factor. This finding may result in extremely low-complexity wireless transceivers in time- division duplex (TDD) cellular systems, where the uplink MUD and downlink MUT can be implemented based on one adaptive signal processing scheme.