-----Original Message-----
From:Rolf Taylor
Sent:2006-03-08 18:05:11
To:radio-tech@broadcast.net
Subject:[RT] ISDN over IP (was: IP Audio and Wireless ISP?)
Kevin krider at ix.netcom.com writes:
Why can't someone make an ISDN emulator for IP? How about a box with and ISDN connection on one side and an Network connection on the other? The network provides 64k (or 128k) of bandwidth and the box provides
everything
>else to make it look like an ISDN circuit to our old ISDN codecs. I need this device...
Kevin,
very good question.
This is NOT a trivial task, but the answer is: this exists, and their ISDN
implementation has been tested with Zephyr and Zephyr Xstream. See:
http://www.patapsco.co.uk/applications/
isdn_over_ip/isdn_ip.htm
and
http://www.patapsco.co.uk/pdfs/ISDN%20Codec%20
over%20IP%20
Application%20Note%20V1.02.pdf
However, this does not address the need for quality of service on the IP
link. This is becoming more available, but is still the exception, not the
rule. If you are using this equipment on an MPLS network (see Mike Tosch's
recent thread) that offers sufficient QoS, it is likely to work very well.
I met recently with this vendor, and they have been doing a lot of testing
with various products, not just codecs.
Our biggest area of concern was
the basic compatibility of the ISDN signaling as we've seen compatibility
problems with some PBXs. I am happy to report there were no compatibility
problems at all.
Their experience with IP services without QoS mechanisms is that if you
use the same ISP on both ends, reliability tends to be decent since in
that case it does not actually hit the public internet. But in this case,
there is no performance guarantees like there would be with IP offering
that guarantee QoS.
Operation over the public internet is possible, but obviously the
reliability under those conditions is the lowest. And if you do have
problems, there is almost nothing you can do to get them remedied.
The real key for the case where there is no guaranteed QoS is to use
coding schemes that are inherently robust to errors (such as G.722 or
ADPCM) or to use one of the more efficient codecs combined various
techniques to conceal lost or corrupted packets.
Zephyr Xstream's AAC has error concealment which is why we recommend using
Layer 2 or Layer 3 when testing suspect lines.
|
It is worth noting that this error concealment works best if you were to stream directly Xstream to Xstream over IP versus using the PacketBand equipment to emulate ISDN, since in the former case we control the packetisation. Since it increases payload, forward error correction generally does not help in these cases. Error concealment and amelioration techniques are no free lunch. They can reduce the audibility of errors (or even make them inaudible) but analogous to lossy codecs (ADPCM, MPEG, etc) the output is not identical to the input.
This is an active area of research. So, QoS will always be desirable in much the same way as it will always be desirable to use a greater bit rates for audio coding whenever possible. Another cool thing about Patapsco's gear is that they can also connect IP to an ISDN network.
So using this technology codecs in the IP world could "call" other IP codecs or they could place calls to ISDN codecs. For a large company with a WAN with QoS this could result in some big savings since on net access would be "free" and off net access could be aggregated to save on ISDN costs. Pretty slick, eh?
Patapsco is the only vendor of this type of equipment that we are aware of. They will be at their dealer's booth (in the TV hall) at NAB 2006. If you will be attending, check with me at our booth and I'll have the booth number for them. They also make a nice ISDN PRI to BRI converter, suitable for aggregation of ISND BRI access, that is similar to the Adtran Atlas series.
Rolf Taylor
Telos Systems
2101 Superior Ave
Cleveland,
OH 44114 216
.241-7225
216-241-4103 (Fax)
http://www.telos-systems.com
|