Skip to content

Nonlinear ML receiver

Roberto Louro Magueta requested to merge develop-UL-nonlinear into develop
  • Fix rho computation;
  • Compute LLR for QPSK, 16QAM and 64QAM for nonlinear ML receiver;
  • Select Nonlinear ML or Linear MMSE receiver.

By default, we are using the Nonlinear ML receiver, except

  • for 256QAM as Nonlinear ML receiver is not implemented for 256QAM;
  • for 64QAM as Nonlinear ML receiver requires more processing time than MMSE, and many machines are not powerful enough.

Results with the nonlinear ML receiver

QPSK, MCS = 2

sudo LD_LIBRARY_PATH=. ./nr_ulsim -n2000 -b14 -I5 -i 0,1 -g B,l -t70 -u 1 -m2 -R106 -r106 -U 1,1,1,2 -W2 -y2 -z2 -s1.7 -S1.7 -P

MMSE: Eff Throughput = 93.45, RX PUSCH time 578.15 us
ML: Eff Throughput 98.48, RX PUSCH time 385.05 us

16QAM, MCS = 16

sudo LD_LIBRARY_PATH=. ./nr_ulsim -n2000 -b14 -I5 -i 0,1 -g C,l -t70 -u 1 -m16 -R106 -r106 -U 1,1,1,2 -W2 -y2 -z2 -s18.7 -S18.7 -P

MMSE: Eff Throughput = 57.83, RX PUSCH time 656.67 us
ML: Eff Throughput 98.15, RX PUSCH time 699.09 us

16QAM, MCS = 16, 4 Rx antennas

sudo LD_LIBRARY_PATH=. ./nr_ulsim -n2000 -b14 -I5 -i 0,1 -g C,l -t70 -u 1 -m16 -R106 -r106 -U 1,1,1,2 -W2 -y2 -z4 -s11.2 -S11.2 -P

MMSE: Eff Throughput 93.98, RX PUSCH time 1118.04 us
ML: Eff Throughput 99.60, RX PUSCH time 1105.34 us

64QAM, MCS = 17

sudo LD_LIBRARY_PATH=. ./nr_ulsim -n2000 -g C,l -m17 -s25 -S25 -W2 -y2 -z2 -P

MMSE: Eff Throughput 87.08, RX PUSCH time 311.89 us
ML: Eff Throughput 97.20, RX PUSCH time 687.86 us

Merge request reports