Presence and IntComm WG

[Jeremy George <>]

  Not so good news and good news.  The not so good news is that I failed
in coming up with a simple chart plotting mos v packet loss for gsm.

  The good news is that I discovered that the ITU has created an 'e-model'
(G.107 I think?) meant to replace mos scores on packet networks.
Very basically, the e-model produces an r-value, the calculation of
which appears complex to my three-semesters-of-calculus-decades-ago eyes.

  Two interesting sources discussing the issues are:

atr.alcatel.de/hefte/00i_1/gb/pdf_gb/06vleegb.pdf
www.aptsec.org/seminar/meeting-2003/policy-forum/documents/INF05_Article-VOIP-QoS-Benchmarks-APT.doc

  The broad relationship of r-value to mos score is:

      r-value   mos
      100-90     5     excellent
       89-80     4     traditional toll quality
       79-70     3     useable, some dissatisfaction expected
       69-60     2     audible, poor
       59-50     1     barely audible, very poor

  GMS-FR (right?  there are actually three flavors of gsm) in ideal
conditions gets an r-value of 74, equivalent to a 3.74 mos score.  What
I couldn't find is how it performs in increasingly degraded circumstances.

  For the SMM trial maybe the best thing is to leave the weather report
as is?  But, I think exploring the e-model would be very useful for
schools considering moving to VoIP in a big way.  Perhaps this would be
something the VoIP ITEC might consider?

- Jeremy

--