|
|
|
Commetrex ...Technology for the New Network
1. Grab a Hammer and See If It Works
2. Are You Bidding on Networx?
3. T.38 Version 0, 1, 2, 3: Should You Care?
4. Not Your Father's UM System
Grab a Hammer and See If It Works
It's taken a long time, but it
appears that gateway and endpoint vendors are finally finding out that a fax
call can't be handled as a voice call. True, most figured out some time ago
that dynamic jitter buffers and G.723.1 didn't work. Now, the link between T.38
and customer satisfaction is becoming clear. And carriers are now deploying
hosted enterprise fax based on fax media servers that terminate T.38 sessions.
The Hammer FX from Empirix, equipped with Commetrex' exclusive Multi-Modal
Terminating Fax (MMTF), is designed to allow them to test their applications.
Hammer FX not only allows you to test voice-based
applications, such as IP PBX, VoIP gateways, and audio conferencing systems. But
with MMTF, Hammer FX can inject T.38 fax transactions into the network and
monitor for proper results.
MMTF combines Commetrex' TerminatingT38, which, for
years, has been the reference standard in the T.38 Interoperability Test Lab,
with Commetrex' fax modems. This allows Hammer FX to send-receive T.38-based
faxes and to perform the same functional tests on gateways and endpoints that
have yet to add support for T.38.
Empirix' Hammer Call Analyzer (HCA) also includes
TerminatingT38. With HCA, users can analyze and decode T.38 transactions over
UDP or TCP with both SIP and H.323 protocols. HCA also provides valuable fax-quality
metrics, including transmission times, quantity of T.38 Indicator and Data
packets, and HDLC events.
To learn more about how Empirix is using Commetrex
technology to Hammer T.38 sessions, go to http://www.empirix.com
or call 1-800-Hammer-IT.
To learn more about MMTF, contact Cliff Schornak at cschornak@commetrex.com
or 770-449-7775 X330
Are You Bidding on Networx? Team
The US General Services
Administration (GSA) has solicited proposals for the "Networx" program
(http://www.gsa.gov/networx) that will provide "comprehensive, best-value
telecommunications and networking services and technical solutions to all
federal agencies. Notice the last phrase: "to all federal agencies."
This is a very big procurement that will probably result in billions of dollars
in business to many telecom equipment, solutions, and service providers. GSA is
looking at this procurement strategically. According to its Website, "Networx
will introduce new technology, new industry partners, and new ways to achieve a
more efficient and effective government."
The bid documents are available in two versions: "Enterprise"
and "Universal." The specification for each is in Section C. Section
C.2.4.5 Internet Facsimile Service (IFS) of each of these specifies a
comprehensive, advanced set of fax facilities. We believe that these
requirements may be uniquely supported by Commetrex' "Open
Telecommunications Framework, A New Architecture for Fax Telephony", which
is fully described at http://www.commetrex.com/whitepapers/OTF2.html.
If you are bidding on Networx, and would like to
increase your bid score, Commetrex can offer you a 100% conformant solution. We've
also explained how OTF does this in a written response to section C.2.4.5. If
you are a qualified bidder, and would like a copy of our response, contact Mike
Coffee at 770-449-7775 X310 or mcoffee@commetrex.com.
|
|
T.38 Version 0, 1, 2, 3: Should You Care?
T.38, the ITU recommendation for
real-time fax over IP, was first determined in June 1998. Since then, there
have been corrections, amendments, and new versions. Hope this orientation
helps you understand the differences.
One of the most interesting had to do with the ASN.1 (Abstract
Syntax Notation) used in T.38. ASN.1 is a notational language that formalizes
the specification of communications protocols, and reduces the amount of
signaling data that must be transmitted relative to text-based protocols. It's
similar to a programming language in that the binary output is precisely
described by the input. So don't make a mistake. If you do and the mistake
makes it through the review process (sometimes they are not obvious) you have a
problem. And that's just what happened in the first-what later become known as
Version 0-T.38. A comma followed by a return and an ellipsis on the next line
is what was omitted.
Of course the typo was discovered, and, in 2002, Version
2 of T.38 corrected the error. But wait! By that time several companies,
including Commetrex, had implemented the ASN.1 specified in the first version.
So gateways that implemented the first and "corrected" version of T.38
(V2) were not interoperable, clearly a bad outcome for the correction. So, in
July 2003, a correction (Corrigendum in ITU speak) was issued. It recommends
that version numbers be a mandatory attribute in T.38 to indicate which ASN.1
syntax is used.
So the responsible committee decided on version numbers
that would allow corresponding gateways to use the same encoding. (If no
version number is provided by a gateway, Version 0 is assumed.) So the
corrected corrected version became Version 2 along with its Corrigendum.
Earlier, Version 1 added TCP; later, version 3 added V.34
support.
The chart below details the versions and the major
changes of each:
| Ver |
Form |
Date |
Changes |
| 0 |
T.38 |
June 98 |
T.38 w/ASN.1 error |
| - |
Corrigendum |
April 99 |
Noted ASN.1 incompatibilities |
| 1 |
T.38 |
Nov 2000 |
TCP and IAF support |
| 2 |
T.38 |
2002 |
Updated Recomendation changing ASN.1 syntax |
| - |
Corrigendum |
July 2003 |
Made version numbers a mandatory attribute. Tied version num. Negotiated
to ASN.1 syntax used. |
| 3 |
T.38 |
July 2004 |
Added V.34 support |
|
Not Your Father's UM System
Many unified/integrated-messaging
vendors define the functional scope of UM to be whatever they provide, typically
voice and e-mail. Those that support fax expand the definition accordingly. Now,
vendors of messaging systems to the mobile operator must go well beyond voice to
support video. Not only is video support required, vendors must support a
plethora of terminal devices: PCs, black phones, 2-, 2.5-, and 3-G mobile phones,
fax terminals, and PDAs. Moreover, video presents vendors with a whole new set
of problems, not the least of which is video transcoding.
MPEG1, MPEG4, and H.263 are typical of the mix. A
message stored in one format may not be supported by the endpoint device
retrieving the message; transcoding is required. Those of us with a voice-transcoding
background may look at video transcoding as a straightforward exercise. But
there's much more to video transcoding than just the large doses of MIPS
required.
While speech codecs encode and decode a stream of speech
data, modern video codecs do more than simply encode the video feed; they "describe"
video scenes. One way a codec reduces bandwidth is to describe the scene,
delineating background and the moving portions, and then only send information
on the changes over time for the moving portion of a scene. A talking head, for
example, may have a static background with only small changes in facial features
over time. Moreover, there are big differences in frame rates, data rates, and
resolution between the standards. All of this must be resolved in the
transcoding process.
But, one approach to go from one encoded format to
another would be to do it by "brute-force." This involves decoding
the stream all the way down to composite video and then re-encode the stream (e.g.
RGB, YUV, YcbCr) into the desired encoded format. Since video codecs are not
lossless, this technique degrades the image quality. It also requires as many
MIPS as the original encode-decode, and we all know about how MIPS-intensive
video is.
A better method is to "map" the scene
descriptions from one format to those of the target format without resorting to
decoding and re-encoding the motion estimation, preserving both video quality
and MIPS. This requires a deep understanding of the source and destination
codec standards. It also means you can't mix and match. A voice transcoder can
handle all combinations of all codecs as long as every transform has a linear
PCM input. But the video transcoding approach described above means each
transcoder supports a single input to a single output.
We will have more to say about video in the future, but
in the meantime, if you're interested in video transcoding and would like to
talk with a leading industry expert, contact Jian Liu, Director, Signal
Processing Technologies, at jliu@commetrex.com.
|