In this paper, we propose a reduced-complexity multi-user MIMO precoding scheme based on sorted-QR decomposition suitable for green applications. In addition, a new power allocation strategy is proposed that approaches optimized BER with low arithmetic complexity. The proposed block-based power allocation strategy allocates adequate power to the deeply-faded spatial pipe, which improves bit-error-rate performance. It also partitions the remaining pipes into several blocks with elements of similar magnitude to reduce complexity by exploiting the matrix property after sorted QR decomposition. Finally, the Tomlinson-Harashima precoding is used to get rid of multiple access interference and inter-antenna interference in advance at transmitters. Simulation results show that our proposed scheme is competitive to block-based geometric mean decomposition systems with much reduced arithmetic complexity. Moreover, the proposed scheme shows its superiority when more users are supported in 12×12 antenna configuration, which can be established in fixed and moderate-size transmitters such as the road side unit in intelligent transportation systems.