search

bastibl / gr-ieee802-11

IEEE 802.11 a/g/p Transceiver
https://wime-project.net/
GNU General Public License v3.0
744 stars 290 forks source link

performance issue on mcs 6 (rate 48Mbps) #375

Closed cloud9477 closed 1 year ago

cloud9477 commented 1 year ago

Hi, I did a simple test on performance (packet delivery ratio vs SNR), the curve of mcs 6 (legacy rate 48 mbps) is weird, I was using the equalizer 0, attached please find my results:

psdu len 114: wime-114

psdu len 264: wime-264

psdu len 564: wime-564

bastibl commented 1 year ago

did you run this with the most recent maint-3.10 branch?

cloud9477 commented 1 year ago

yes

bastibl commented 1 year ago

Since this is the only MCS with a code rate of 2/3, I'd assume that there is an issue with certain frame sizes for this MCS. You could (1) use the generic decoder and see if it shows the same behavior or (2) look into rate-specific parts of the code, e.g., https://github.com/bastibl/gr-ieee802-11/blob/maint-3.8/lib/viterbi_decoder/base.cc#L39-L59

cloud9477 commented 1 year ago

OK, I'll check and see if i could fix it.

bastibl commented 1 year ago

Did you find the reason for the strange behavior?

cloud9477 commented 1 year ago

Did you find the reason for the strange behavior?

Not yet, I try to decode the bits from the equalizer with my decoder, the performance is similar, and also I check the equalizer and seems no issue there. I guess maybe it's just the hard bits lost too much information.