From: jim bell <jimbell@pacifier.com>
Date: Sun, 4 Aug 1996 09:51:50 +0800
To: "Chris Adams" <adamsc@io-online.com>
Subject: Digital Telephony costs $2
Message-ID: <199608040010.RAA21627@mail.pacifier.com>
MIME-Version: 1.0
Content-Type: text/plain


At 02:52 AM 8/3/96 -0800, Chris Adams wrote:
>On 3 Aug 96 01:16:48 -0800, jimbell@pacifier.com wrote:
>
>> I propose that the better way of implementing it, rather than going through 
>>a sound card, is for modem manufacturers to built an  new modem with an 
>>extra telephone connection (perhaps the same physical connector that's 
>>currently used for the telephone handset) which goes to an ordinary 
>>telephone and does the audio A/D and D/A conversion, as well as the data 
>>compression/data expansion function that will be necessary.  The latter 
>>function would be done by an extra DSP on this modem/Internet telephone 
card. 
>
>If you were so inclined, you could implement the whole thing for MWave
>modems.  They are fast enough to handle 28.8k and sound card functions at
>the same time off of a single DSP. 

That seems a bit difficult to believe.  I get the impression that 
implementing a 28.8Kbps+ modem pretty much uses up the capability of a 
near-state-of-the-art DSP chip.  Further, recently an item appeared on CP 
concerning a new voice-compression standard that was claimed to put 
good-quality voice into a 2400 bps stream.   Each function, coding and 
decoding, was claimed to occupy about (don't recall the exact figures) a 
little over half the capability of a TI 32025 DSP chip, which admittedly is 
an older unit.  Assuming full-duplex is desired (and that's the purpose of 
this exercise) you'd need the full resources of something with greater 
'ooomph' than a 32025 just for the coding/decoding.

Sure,  it may not be necessary to compress voice audio all the way down to 
2400 bps, since the current modem standards allow 28.8kbps and beyond, but I 
suggest that decreasing net traffic by a factor of 12 (28.8k to 2.4k) is a 
desirable goal.  Remember, in the long term "everybody" will be using 
Internet telephone. (And no doubt you've noticed that high-volume hardware 
gets cheap, really fast.  Putting in a second DSP for compression/encryption 
won't increase the costs all that much.)  Leaving the encryption in hardware 
would improve exportability, at least from a legal/ITAR standpoint. 


 While eventually full-fiber-capacity Internet will be able to increase the 
capacity to "unlimited" levels, in the meantime the capacity is limited (by 
switches if nothing else) and going the extra mile to limit Internet 
telephone's impact on the national net would be better.


>With the right drivers you could use
>the telephone/speaker/mic jacks that are on most of the integrated cards.
>  Also, they have a standard realtime OS with most of the functions being
>portable across cards as well, so you'd have to do very little work to
>support other Mwave cards.

The reason I think a system I've described has a future is that modem 
manufacturers have a PROBLEM.  Their problem is that they've pretty much run 
out of room to improve the bit-pushing through a 3 KHz bandwidth.  Sure, 
they can focus their attention on cable modems or ISDN units or other toys, 
but the market for such beasts won't develop for a few years.  They'll 
pretty much be stuck doing an occasional upgrade, or selling into new 
computers, but that will slow down.  What they'd like to have is a new 
function that "everybody" wants to have, and allowing people to bypass LD 
telephone charges is a powerful motivating factor to get people to upgrade 
their modems.



Jim Bell
jimbell@pacifier.com