Mike:
Mike the answer is call processing.
It is hard to tell from the MOT press release, but I will guess that they are talking about a media
gateway with a TR303 interface to a CO class 5 switch.
My guess is it will accept DISCOS 1.1 VOIP packets and signaling (based on MGCP), and
convert it to TDM (digital circuit) channels and TR303 (ISDN like) signaling messages. The IP
doesn't run through the switch- TDM does, where it is switched to an outgoing trunk and into the
network. You have to convert to TDM at some point, and it doesn't much matter if it is at the
gateway or the network edge. And the signaling in the CO is sent out over a T1 SS7 trunk
interface at the other end of the CO for a non-local call.
For those blessedly unaware of telecom signaling, let me explain, if you are interested.
A typical central office (CO) has a line side (to you, the subscriber) and a trunk side (to them, the
network: people like ATT, MCI, or to their own ILEC's tandem switch)
The signaling on the line side to your analog phone is DC voltage-shifts for onhook, off hook, and
hookflash, and dial digits), tones, and ringing at 48 volts. You run a twisted pair line from your
house to the CO and an analog line card in the CO switch converts the voice to digital , and the
signaling to some internal digital messaging in the switch. And visa versa.
Now along came an idea: the digital phone. The proposers of ISDN came up with the idea of
having a subscriber line side phone with two digital bearer (or voice) channels, and a digital signaling
channel. Instead of "on hook" "off hook", and hookflash as voltage changes, beep beep tones and
ringing voltage, and other legacy analog signaling, you had advanced new ISDN digital messages for
the signaling. It is now as modern as your digital watch- exempt that phones cost $400 and the
two offered bearer channels cost more than buying two analog lines, and the quality was the same
as analog, so no one bought them.
Note that legacy old analogy phones are still here in their $20 to $80 glory. And ISDN digital
phones are, well, more legacy than legacy?
So ISDN failed, but the CO's had spent 100 millions of $ to upgrade their switches from old analog
to new digital subscriber side signaling.
Now the problem with ISDN was that 2 channels were not enough to be cost effective. But what if
you had, say, 100 channels using the one signaling channel. So a new idea was born- why not take
the signaling messaging, and put it on a dedicated channel of a T1 or T3 or something. Also reserve
a second channel for 'management', so the CO and ILEC could control the device at the other end.
You now can have a very large 'subscriber side' interface with many TDM channels. This is cost
effective and more easily managed than other solutions, so everyone is happy. The subscriber get
since ISDN features at a lower cost, and they let the phone company manage the whole thing.
But why would MOT, or anyone else what to use nasty old legacy circuit switch COs?
Well, the CO's contain about 400 staff years of code- at a minimum- that supports call processing.
And another 100 staff years each year for maintenance. Until you build a complex set of software
running various servers (a four year effort several start ups are attempting), you can't make a phone
call. Which means you can't sell your lovely high margin high volume VOIP equipment. So if you
make a TR303 interface to a CO, and keep your media gateway equipment "dumb"- meaning it is
only a signaling protocol and voice IP to TDM converter- you can reuse COs.
This is the way the roll out will happen, in my opinion:
1. In the next two years, Cable equipment providers will sell media gateways which take DOCSIS
1.1 in, and put T1/TDM with TR303 signaling out. And visa versa. The CO will provide the call
processing, billing, management, and other necessary features like LNP, CALEA, 800/900 calls,
etc.. Lucent and NT will sell more COs.
2. In a few more years, Since the call model of a tandem switch (Class 4 switch) is much simpler
and a CO (Class 5 switch), various small start ups will offer a set of software and UNIX based
servers that implement a tandem call model using an MGCP like protocol.
Now remember when we went from analog subscriber side signaling to ISDN digital- well another
bunch of people got a better idea, and invented various forms of IP based signaling, one of which
came out of the IETF (the folks who gave us the internet) and they called it MGCP (media gateway
control protocol). So you can use this instead of analog or ISDN to manage subscriber side voice
devices. However, you will still need the media gateway to convert from Ip voice packets to TDM
to get to the network, but the signaling will be from DOCSIS 1.1 to SS7 directly, without a CO
involved. So for long distance calls, we get rid of the legacy CO.
3. After four or more years, there will be pressure to keep internal DOCSIS network calls (the local
free calls) on the network, and not go out to a TDM switch. So someone will code a working CO
Class 5 call model, and it will be deployed. Then the CO will be replaced by a set of things called
'media gateway servers'. Basically, you are separating the 'switching' hardware from the software
control.
The real problem with 2 and 3 is not technology, it is time, and also money. And money is more
important. Lucent, Nortel, Alcatel, etc. have to build hardware and software. 80% of development
cost is software, but 80% of revenue is from a per line hardware cost. If you don't bundle the
hardware with the software, how do you charge for server software, to recover the 400 staff years
of effort and a 50% margin? And how to you continue to pay the 100 staggers for their
maintenance of software? These numbers are low, by the way.
This is why the TR303 solution will be around for five years, at least. And Lucent/NT, etc. will buy
these small companies, bundle the hardware and software, and hey, we will be back where we
started.
The real revolution will be phase 4, when the subscriber side contains the call processing- that is,
your class 5 call processing is on your computer rather like your browser is on your computer. But
this is long long term, and I have no idea of a business model that will work to deploy and support a
home based CO, while paying the salaries and stock options of 100 programmers, 100 help desk
technicians, and probably 100 managment/admin staff. Not to mention how the thing is managed.
Once again, technology is not the problem, the business case is. |
