Loading ...
Sorry, an error occurred while loading the content.

3287Re: [V4Protocol] Re: [WINMOR] Test version 1.4.7.5 of WINMOR TNC posted on Yahoo and Winlink User FTP

Expand Messages
  • Rick Muething
    May 15 6:54 PM
    • 0 Attachment
      Finn,
       
      There is less to be gained by using QAM8 on the 4PSK mode.  There is already pretty heavy FEC. (code rate = 1/2) before RS.  Though it probably won’t be too hard to test it. First I have to find the bug in the QAM16 implementation....it works but it is not correcting as it should!
       
       
      Re MFSK16.  You have to decide what do you want.  If you want the absolute robustness you use wide band, very high redundancy and maximum power. MFSK16 takes up a lot of room for its throughput but it is robust because of all the redundancy.  But a narrower mode like QAM8 that is carrying say 2 FEC bits/data bit (code rate = 1/3) can be very robust since it puts all the power in a smaller bandwidth (higher S/N).   The biggest problem with the QAM modes will come with very strong multipath though lower baud rates help this.
       
      I would like to include UTF8 capability (pure binary) in a mode but it does slow down the text  (8 or 16 data bits vs. 7 for ASCII ).
       
      Rick
       
       
       
      From: la7um
      Sent: Wednesday, May 15, 2013 8:56 PM
      Subject: [V4Protocol] Re: [WINMOR] Test version 1.4.7.5 of WINMOR TNC posted on Yahoo and Winlink User FTP
       
       

      Rick. How robust would this be compared to MFSK16 used as ARQ? And what about UTF8 ?. I am searcing for a commom denomintator for Aurora net control.

      4PSK changed to 8QAM in WINMOR seem the next step regardless of BW modes.

      THEN rethink about when gearing from one BW to another in WINMOR.

      For QSO SPEED, it seems for me that the most robust mode possible at any given time seems of interest.

      After sun burst last day it was mandatory staying LOW in frequency, and in addition forget about all wide bands. Stick to 500Hz BW for making the tests in new WINMOR regardless of distance to my local server.

      Finn/LA7UM

      --- In mailto:V4Protocol%40yahoogroups.com, "Rick Muething" <rmuething@...> wrote:

      >
      > Jakob,
      >
      > Well I hadn’t thought about it much...I’m pretty swamped. But it
      would be possible to build something like this:
      >
      > Replace the
      4FSK modulation in V4 with 8QAM. The current 4FSK carries one FEC bit and one user bit per symbol. 8QAM could carry 3 bits/symbol would be used something like:
      > 1 user bit per symbol (46.875 baud give about 100 Hz of bandwidth
      and about 351 bytes/minute or about 58 words/min (probably closer to 45 WPM after ARQ overhead). That is fast enough for manual typing. The signal would be about 50% wider than PSK31
      >
      > 2 FEC bits per symbol one being the
      Viterbi Trellis code and one being the layered Reed Solomon. That should be very robust in all but the worst extreme multipath propagation. Layering two codes is very productive as they complement each other (Viterbi is near optimum for random errors but not good for bursts. Reed Solomon is excellent for bursts)
      >
      > I may have some time this summer to play with that while I am
      on our RV trip out west to Wyoming for 2 months.
      >
      > Your comments
      welcome.
      >
      > Rick, KN6KB
      >
      >
      > From: dk3cw@...
      > Sent: Wednesday, May 15, 2013 3:15 PM
      > To:
      href="mailto:WINMOR%40yahoogroups.com">mailto:WINMOR%40yahoogroups.com ; mailto:Winlink_Programs_Group%40yahoogroups.com ; mailto:V4Protocol%40yahoogroups.com
      > Subject: [V4Protocol] RE: [WINMOR] Test version 1.4.7.5 of WINMOR TNC
      posted on Yahoo and Winlink User FTP
      >
      >
      >
      >
      Rick,
      >
      > thanks for the improvements to the WINMOR protocol. Just
      tested with RMS DB0ZAV, who is still using version 1.4.7.3. Works fine!
      >
      >
      >
      > Is there an implementation of 32QAM into the V4
      Soundcard TNC planned?
      >
      >
      >
      >
      > vy 73 de
      Jakob DK3CW
      >
      > EmComm liaison officer Frankfurt / Main
      >
      ARDF-unit Hessen
      > www.dk3cw.de
      >
      > When All Else Fails, Ham
      Radio Works
      >
      >
      >
      > -----Ursprüngliche
      Nachricht-----
      > Von: "Rick Muething" [rmuething@...]
      > Gesendet:
      Mi. 15.05.2013 20:56
      > An:
      href="mailto:Winlink_Programs_Group%40yahoogroups.com">mailto:Winlink_Programs_Group%40yahoogroups.com, mailto:WINMOR%40yahoogroups.com, mailto:MARS_Winlink%40yahoogroups.com
      >
      Betreff: [WINMOR] Test version 1.4.7.5 of WINMOR TNC posted on Yahoo and Winlink User FTP
      >
      >
      >
      >
      >
      > All,
      >
      > Doing continued refinement of the 32QAM mode. This version (which
      should be backward compatible with both the 1.4.7.3 and earlier versions) now supports 16QAM in place of 8PSK and 32QAM in place of 16PSK. This should provide improved robustness when connecting to a similar version. I am running this new version with additional logging on my RMS Trimode and would welcome test connections. To activate these higher speed modes (in either 2 carrier or 8 carrier channels) it is necessary to have some fairly large (say 5+ k) messages to insure it tries these higher speed modulations. You should enable the WINMOR TNC debug log (On the WINMOR TNC form Help, basic setup, check Enable WM TNC debug log, Update ini) as this will capture some internal measurements of how well the 16 or 32 QAM modes are working.
      >
      > For those that have
      asked (or for anyone who really cares! ) QAM means Quadrature Amplitude Modulation where both the carrier amplitude and phase are changed (as opposed to PSK (Phase Shift Keying) where only the phase is modulated. This permits sending one extra bit (two amplitude levels) per symbol and this extra FEC bit is used to improve the robustness of these higher modulation modes. You will see this as rings of symbols (inner and outer rings). If the signals are very good you can make out the individual symbol clusters (16 for 16QAM, 32 for 32QAM) but most of the time they are so smeared (by propagation) that they look more like fuzzy concentric rings. The FEC bits help ?recover? these smeared symbols to yield good decoding.
      >
      > I have posted 1.4.7.5 on both the WINMOR Yahoo
      group and the Winlink WINMOR TNC FTP ftp://autoupdate.winlink.org/WINMOR%20TNC/
      >
      > 73,
      >
      > Rick Muething, KN6KB
      >
      >
      >
      >
      >
      > -----Ursprüngliche Nachricht Ende-----
      >
      >
      >
      > ---
      >
      > Alle Postfächer an einem Ort.
      Jetzt wechseln und E-Mail-Adresse mitnehmen! Rundum glücklich mit freenetMail
      >

    • Show all 11 messages in this topic