On 7/14/2010 1:29 PM, Jim Lux wrote:

There's a whole raft of issues..

One is that cellphones aren't analog FM anymore, so they suffer from the
"digital cliff" problem.. If the Eb/No is good enough that the FEC
works, then it works great, but a dB or so worse, and it doesn't work at
all.

This is so different from most people's practical experience with
communications that it is disconcerting. E.g. if you walk away from me
while talking, your voice gets fainter and the SNR gradually drops, but
it's not like you've walked out of the "cone of silence" where it goes
from fully intelligible to not at all intelligible in milliseconds.

And, the fact that this happens on a packet by packet basis means that
at the threshold, it "stutters", which is also non-intuitive and
non-physical.


And then, the cellphone companies have managed to convince us that
really crummy sound is ok, even at the full bit rate of 8kpbs. The
voice codecs are impressive, but face it, it still doesn't sound like
56kbps u-law or 25 kHz NBFM, what used to be called "toll-quality"

And then there's the PSTN ala 2010. We have been discovering a lot of
new things about that that we suspect even the carriers haven't noticed,
or at least appreciated. And it's all started showing up in the last
year. Like average LD call completion failure has been 4% for 10 years.
Yup it's actually that high. Except this year it's 6%. And possibly
still climbing. Can't tell you the test conditions (NDA needed), but
it's millions of calls per year.

One of the first odd things we noticed was the occasional location code
3 from the supposed destination. That one is now obvious, but the first
time you see it it's very perplexing.

And more things are showing up, mostly disturbing if you expect the PSTN
to remain the relatively predictable thing that it's been during the
SS7/TDM/ISDN era. So much is changing we are making a new training
program for our analysts on how to recognize what's now going on versus
what used to go on.

tom
K0TAR