[VoiceOps] T38 over Sip
Alex Balashov
abalashov at evaristesys.com
Fri Aug 28 09:25:10 EDT 2009
Junaid Hattia wrote:
> if a single page is faxed it works!....if multiple pages are sent, only one
> and a half pages of say three pages are received
Yep, well, that's generally where the other problems I alluded to crop
up symptomatically. This is a complex - and, even more importantly in
the case of T.38 and anything fax-related, a multi-vendor - call path.
1) Where is the fax being sent from?
a) Is it a Super G3 fax machine?
b) Have you ruled out the possibility of interference and/or
poor PSTN service quality on the sending side?
c) Does wherever you're sending from consistently work fine
if you send to a conventional fax machine on your side?
2) Have you verified that the switch to T.38/UDPTL actually takes place,
with a packet capture? Or does it stay G.711?
It may be that the data is being exchanged using G.711u analog
pass-through, which is the mode in which things stay if T.38 cannot be
negotiated between all the media endpoints, and/or if fax tones are not
properly detected because the sender is SG3, and/or similar situations.
Analog pass-through will work, to some extent, but not over anything
less than a very, very well-engineered IP network from a QoS standpoint.
Even then, with all the media traversal steps involved here, I don't
know that the reliability issue can be overcome for analog pass-through.
T.30 is exponentially more sensitive to jitter and even the slightest
bit of packet loss than voice; a little QoS issue will cause a faint
(and perhaps not audible) crackle in speech, but it'll cause bits to be
flipped or missing in modulated digital data. :-) And fax was built
for the PSTN and synchronous transport, not at all designed to be
resilient to this packetisation business.
-- Alex
--
Alex Balashov - Principal
Evariste Systems
Web : http://www.evaristesys.com/
Tel : (+1) (678) 954-0670
Direct : (+1) (678) 954-0671
More information about the VoiceOps
mailing list