QAM16 C port

[drowe67]

Following on from drowe67/codec2-dev#130, and part of ARDC grant WP6000

./src/freedv_data_raw_tx --bursts 100 --testframes 1 qam16c2 /dev/zero - | ./src/ch - - --No -35 -f 20 --mpp | ./src/freedv_data_raw_rx --testframes qam16c2 - /dev/null

Results, PER about 0.1 at 15dB MPP:

ch: Fs: 8000 NodB: -35.00 foff: 20.00 Hz fading: 2 nhfdelay: 16 clip: 32767.00 ssbfilt: 1 complexout: 0
payload bytes_per_modem_frame: 1213 Adjusting testframes to equal framesperburst*bursts

payload bytes_per_modem_frame: 1213
mark:space: 0.94 SNR offset: -0.26
ch: SNR3k(dB):    15.69  C/No....:    50.46
ch: peak.....: 16799.86  RMS.....:  5930.03   CPAPR.....:  9.04 
ch: Nsamples.:  2801600  clipped.:     0.00%  OutClipped:  0.00%
modembufs:   3840 bytes: 111596 Frms.:    92 SNRAv:  2.70
BER......: 0.0517 Tbits: 1506600 Terrs: 77885
Coded BER: 0.0011 Tbits: 903960 Terrs:  1006
Coded FER: 0.0108 Tfrms:    93 Tfers:     1

TODO

• working SNR estimator
• see if we can improve PAPR with clipper
• Update snr_curves.sh, might be time to split that into a few graphs
• Octave auto-magically works out #data bits in FEC codeword (like C)

[drowe67]

@DJ2LS ☝️

[DJ2LS]

@drowe67 Does this already work with feature request 001?

[drowe67]

@drowe67 Does this already work with feature request 001?

Yes I think it should be supported. The QAM logic kicks in when bps=4