LDPC on BSC channel - zero BE, FE

[ronzhin-dmitry]

Hello!

I've installed AFF3CT via cmake, no problems were encountered.
However, when I try to simulate LDPC over BSC using provided example matrix, I get zero BE. Here is the command I launch:
./aff3ct -C "LDPC" --dec-h-path conf/dec/LDPC/AR4JA_4096_8192.qc -m 0.5 -M 0.5 --sim-stop-time 120 --chn-type BSC

Here is the output:

# ----------------------------------------------------
# ---- A FAST FORWARD ERROR CORRECTION TOOLBOX >> ----
# ----------------------------------------------------
# Parameters:
# * Simulation ------------------------------------
#    ** Type                      = BFER
#    ** Type of bits              = int32
#    ** Type of reals             = float32
#    ** Date (UTC)                = 2023-02-17 12:09:14
#    ** Git version               = v3.0.2
#    ** Code type (C)             = LDPC
#    ** Noise range               = 0.5 -> 0.5 dB
#    ** Noise type (E)            = EP
#    ** Seed                      = 0
#    ** Statistics                = off
#    ** Debug mode                = off
#    ** Inter frame level         = 1
#    ** Multi-threading (t)       = 8 thread(s)
#    ** Coset approach (c)        = no
#    ** Coded monitoring          = no
#    ** Bad frames tracking       = off
#    ** Bad frames replay         = off
#    ** Bit rate                  = 0.4 (2/5)
# * Source ----------------------------------------
#    ** Type                      = AZCW
#    ** Implementation            = STD
#    ** Info. bits (K_info)       = 4096
# * Codec -----------------------------------------
#    ** Type                      = LDPC
#    ** Info. bits (K)            = 4096
#    ** Codeword size (N_cw)      = 10240
#    ** Frame size (N)            = 10240
#    ** Code rate                 = 0.4 (2/5)
# * Encoder ---------------------------------------
#    ** Type                      = AZCW
#    ** Systematic                = yes
# * Decoder ---------------------------------------
#    ** Type (D)                  = BP_FLOODING
#    ** Implementation            = SPA
#    ** Systematic                = yes
#    ** H matrix path             = /mnt/d/Work/GitHub/aff3ct/build/bin/../../conf/dec/LDPC/AR4JA_4096_8192.qc
#    ** H matrix reordering       = NONE
#    ** Num. of iterations (i)    = 10
#    ** Stop criterion (syndrome) = on
#    ** Stop criterion depth      = 1
# * Modem -----------------------------------------
#    ** Type                      = BPSK
#    ** Implementation            = STD
#    ** Bits per symbol           = 1
#    ** Sigma square              = on
# * Channel ---------------------------------------
#    ** Type                      = BSC
#    ** Implementation            = STD
#    ** Complex                   = off
#    ** Add users                 = off
# * Monitor ---------------------------------------
#    ** Lazy reduction            = off
#    ** Frame error count (e)     = 100
#    ** Compute mutual info       = no
# * Terminal --------------------------------------
#    ** Enabled                   = yes
#    ** Frequency (ms)            = 500
#
# The simulation is running...
# -----------------||------------------------------------------------------||---------------------
# Event Probability||   Bit Error Rate (BER) and Frame Error Rate (FER)    ||  Global throughput
#       (EP)       ||                                                      ||  and elapsed time
# -----------------||------------------------------------------------------||---------------------
# -----------------||----------|----------|----------|----------|----------||----------|----------
#               EP ||      FRA |       BE |       FE |      BER |      FER ||  SIM_THR |    ET/RT
#                  ||          |          |          |          |          ||   (Mb/s) | (hhmmss)
# -----------------||----------|----------|----------|----------|----------||----------|----------
            0.5000 ||   326351 |        0 |        0 | 7.48e-10 | 3.06e-06 ||   11.139 | 00h02'00
# End of the simulation.

Using -m=-M=1.0 and -m=-M=0.0 changes nothing (although I supposed 1.0 means constant error in BSC).

Do I understand something wrong, what is the reason for zero BE, FE?
Thank you in advance.

[Alex-Xiaoyu-Ai]

Have you tried a larger code rate (say 0.75) instead of 0.4? There is a chance that for the rate-0.4 code used in your simulation is capable of correcting all the errors at a noise level of 0.5dB.