Design of low-latency uplink MAC scheduling for Massive MIMO-OFDM systems

Massive MIMO is a promising technology for 5G due to its huge extended multi-user spatial multiplexing and beamforming capability, but its high complexity makes the associated MAC scheduling suffer from long processing time in practice. To solve such problem, this paper designs a low-latency uplink MAC scheduling scheme for Massive MIMO-OFDM systems. Specifically, first, a framework enabling processing parallelization is put forward in which the scheduling for multiple resource blocks (RB) are executed concurrently with the common pre-processing and post-processing. Second, a power-based weight is introduced into the per-RB scheduling, by which the link adaptation could be completed simultaneously with per-RB processing, no longer requiring reduplicate post-scheduling processing. Third, a low-complexity calculation method is applied to the per-layer recursive user selection algorithm for each RB, in which the matrix inversion is substituted by the recursion with vector inner-productions based on Hermitian Matrix Blockwise Inversion Lemma. Finally, performance simulation results demonstrate that the designed MAC scheduling scheme outperforms random scheduling in terms of both cell sum throughput and cell edge throughput.