At 10:57 AM +0200 8/19/02, TOMSON ERIC wrote:
To exceed that limit, we need to replace POTS with (e.g.) ISDN
(2x64kbps), CATV (256-512kbps), ADSL (1Mbps)...
The local loop - house to CO - can support more than 56K - that's how
DSL works. The problem is in the trunking of voice data. E.g., for
a T-1, a DS-0 is sampled for 7 bits of value 8000 times a second (and
1 bit of signalling). 24 DS-0's are trunked on to a DS-1/T-1 frame.
There's also a signalling bit for the frame.
(7+1)bits x 24 + 1 = 193bits @ 8000/sec = 1,544,000 = 1.544 Mbps (A EE's "M")
For a single DS-0:
7 bits * 8000/sec = 56,000 bps - that's where the magic number comes from
The 8000 times a second comes from the Shannon's (or was it
Nyquist's?) research that found that sampling at 2 x the bandwidth
(bandpass - the range of frequencies passing) was optimal. I.e., the
local loop is assumed to have a 4k hertz range (by the sampling
systems).
With "other limiting factors," such as fiber termination problems, 4k
isn't always reached, and is why one rarely (if ever) sees 56k bps on
a DS-0.
SONET frames (I think, it's been a while) also run at 8000 times per
second, my guess has been that this is to keep some semblance of
similarity between POTS and SONET.
With a sampling of 8000 times a second, if I'm not mistaken, the
maximum baud rate is 4000 symbols per second. By making a symbol
mean more than one bit, you can go faster. Assuming V.90 is 2400
baud, then to reach 56k you'd need to have about 24 (really 32)
symbols in your alphabet, which is 5 bits/symbol. Not having read
the V.90 specs, they may have used a higher number of bits/symbol and
left some symbols unused for error control.
None of this has anything to do with compression. That comes in to
play above the 56k limit. What's put into the frame is compressed
ahead of time, the result of the compression is limited to 56k by the
trunk sampling.
--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis +1-703-227-9854
ARIN Research Engineer