System-level performance for MMSE multiuser detection in asynchronous mixed service UMTS networks

The main objective of this paper is to investigate asynchronous linear multiuser detection (MUD) in mixed service UMTS networks in terms of system capacity as well as power reduction at mobile stations. This especially requires us to theoretically describe the interaction of linear multiuser detection with radio resource management concepts like power control and load control for asynchronous multirate transmissions. For system capacity assessments we apply a two-stage power control multiuser detection algorithm where the users are detected by a minimum mean square error (MMSE) block detection. By detailed system level simulations we evaluate the UMTS load control parameter noise rise and also show corresponding system capacity bounds for different service mixes. It is important to note that for typical UMTS services the signal-to-noise operating points of the MMSE detector are very low (less than 0 dB). These low SINR operating points severely limit multiuser detection performance gains. Compared to simple single-user matched filtering the MMSE leads to fairly modest capacity gains of up to 60% dependent on the service mix assumed.