Answer: The Polycom VTX 1000 conference phone.
The thing that makes this fact so curious is not immediately obvious. The VTX 1000 is not a SIP device, nor even IP capable. Like it’s closest relative the Polycom SoundStation 2, it’s designed to connect to a plain old analog phone line (a.k.a. POTS, the PSTN).
Of course, the common wisdom is that you just can’t have wideband telephony over the PSTN. Yet the VTX 1000, circa 2003, delivers wideband conference calls so it seems that assertion is not strictly true. Understanding this is in a little more depth would seem like a good thing. Happily, Polycom published a white paper describing the internals of the VTX 1000.
The VTX 1000 was originally brought to my attention by David Frankel, CEO of the ZipDX wideband conference service. It is essentially a dual-mode device supporting both normal G.711 conference calls like its predecessors, and wideband calls when it determines that the far-end device is another VTX 1000.
The VTX 1000 behaves in some ways much like a FAX machine. Internally, it’s analog line interface is basically a v.34 modem. When it places an outgoing call it sends a sequence of tones, not unlike FAX tones, intended to help the answering devices identify that the calling coming from a VTX 1000.
When a call is answered it listens for these tones. If they are not heard then it knows that it’s just received a normal phone call and it should act appropriately to handle that call as an analog signal source.
If it does hear the calling party declaring itself to be another VTX 1000 then it handles the calls completely differently. Instead of setting up an analog channel it acts like two modems or FAX machines would, and establishes a point-to-point digital channel. This is like setting up an ad hoc network connection over a pair of v.34 modems. You end up with a data channel capable of up to 33 kbps.
With suitable line conditions a pair of VTX 1000 units can pass calls encoded with a wideband codec like G.722.2, which requires from 12.65 kbit/s to 23.85 kbit/s. While the codec used might be based upon recognized telecom standards, the rest of the call setup process can only be viewed a proprietary protocol. Even so, the system delivers real, wideband calls offering frequency response of 80 Hz to 7 KHz….but only between like devices.
Next question: What constitutes “like devices?”
Excellent question! A device like this is meaningless unless there are other like it elsewhere with which it might connect. Of course, other VTX 1000s would work, also the Polycom VSX 8000 Series video conferencing systems, and finally their Vortex EF2280 Integrated Automatic Mic / Matrix Mixer. This last device being the sort more typically found in larger meeting room installations.
The idea that we might be able to achieve wideband over the PSTN is more than a little heretical in some circles. I was reminded of it by long time VoIP blogger Aswath Rao. he asserted in a tweet not long ago that;
“As you well know SIP URI and IP are not necessary for WBCodec. It can be done in PSTN without involving the carriers”
and further
“…by deploying new devices just like fax caught on, if the market feels the need.”
What Aswath is getting at is the idea that wideband calling doesn’t necessarily require wholly bypassing the PSTN. The very existence of the VTX 1000 supports this belief.
His assertion is also supported by the historical existence of the STU-III secure phones which he referenced in a blog post on March 7, 2009. Developed by the NSA in the late 1980s these phones employ strong encryption to secure calls made over the PSTN. When both end-points are encryption capable the callers can “go secure” and know that the content of their call is traveling over a secure channel.
So in pursuing wideband telephony using SIP over IP networks, essentially bypassing the PSTN, are we throwing out the baby with the bath water?
David Frankel points out that there is more to be learned by examining the typical end-users experience with the VTX 1000. A long-time veteran of the telecom industry David has been promoting wideband for longer than most people. He asserts that while the VTX 1000 was wideband capable most users rarely, and some possibly never experienced wideband calling because of the typical manner of its use.
Lets presume that a company of a significant size buys a number of VTX 1000 systems so that there’s one installed in each of its conference rooms at various locations. If any two offices are going to have a conference call they can make a point-to-point call between the two conference rooms and a wideband call will result. Happy times.
However, if there is a requirement for additional parties on the call they’ll need to use a conference bridge of some sort. Even though the VTX 1000 has been available for six years extremely few conference bridges* support the proprietary VTX 1000 protocol. Because of this fact it’s most likely that everyone on that call would have a narrowband conference experience. The investment in the wideband-capable conference phones will yield the best possible G.711 conference calls, but not necessarily wideband calls.
In fact, in many cases people use a conference bridge even when only connecting two locations. Being able to circulate just one phone number, that of the conference bridge, makes organizing the call easier. So everyone calls the bridge rather than establishing a point-to-point call between conference rooms. In any such competition convenience wins over wideband calling.
There are several lessons in this that I can see. The first is that wideband telephony can be delivered over the PSTN. It’s effectively tunneling through the PSTN. Many possible issues make it less than ideal, but it can be done.
The second is that for any new telephony technology to survive it needs to have an ecosystem developed around it. It needs to be supported by a diversity of devices and services. The lack of that ecosystem dooms it be a short-lived or niche solution. This puts proprietary systems at a significant disadvantage.
The last and perhaps larger lesson is that convenience trumps quality, every time. If we want to deliver the superior call quality of wideband calling, or any other new telephony service, we need to focus on making it just as convenient as dialing a desk phone. It needs to be transparent to the end-user.
Many thanks to Aswath Rao for inspiration through questioning my assumptions, and David Frankel for openly answering the questions of an inquisitive blogger.