The lattice-reduction (LR)-aided successive interference cancellation (SIC) is a well-known detection technique achieving near-maximum-likelihood (ML) performance in the multiuser system. However, the LR algorithm and the computation of sequential nulling vectors still incur considerable computational complexity overhead for a large number of mobile stations (MSs). To overcome it, we propose a decoupled LR-aided SIC for clustered MSs, in which the entire system is linearly decoupled into multiple lower dimensional subsystems before the LR-aided SIC is applied. In addition, efficient MS grouping algorithms are proposed to optimize the detection performance of the decoupled LR-aided SIC. The proposed LR-aided SIC with linear preprocessing exhibits a full diversity order, compared with the conventional LR-aided SIC, with considerable computational savings. Finally, to further improve the system performance, a nested LR-aided SIC with preprocessing is also developed.