And we thought the text part of the Starr Report would be bad

Mailman List Mirror (Read Only)
1

Miquel van Smoorenburg writes:

> There is more to this than meets the eye -- 28.8K is asynchronous
> and has start and stop bits for every byte, so there are a maximum
> of 2880 bytes/sec available over 28.8K. Then there is the issue

Almost every modem supports V42 error correction, which makes the modems
speak a sort of synchronous with each other (actually data is transmitted
in blocks with a start-of-block and end-of-block marker, and a checksum).
That gets you 8 bits in a bith minus some negligeble V42 overhead.
So an 28k8 modem can actually transfer almost 3.6 Kbytes/sec.
Because of the block-oriented approach you do get a bit higher latency
on interactive connections, which is why gamers often turn of V42.

  As long as we are being specific about this, I omitted data
  compression too, which has a real effect on the data transfer rate
  as well. I also omitted 56K modems, ISDN dial-up, and other aspects
  of dial-up communications (like MPP). The point I was trying to
  make was not to disassemble 28.8K modem protocols, but to point
  out the fallacy in believing that a T1 line can only support
  53 modems.

  As far as the start and stop bit issues go, they are controlled in
  the UART on the PC serial port, not the modem. Regardless of how
  the modem encodes and transmits data over the wire, the start and
  stop bits remain in the asynchronous serial protocol on PCs -- years
  ago, PC terminal programs used to let you control the number of start
  and stop bits you used to communicate with a device. The fact that
  DCE/DTE rates are usually higher that the actual speed of the modem
  connection negates some of the effect of start and stop bits, but
  they do exist.

  For modem to modem communications, start and stop bits were used
  up through 2400 baud when one bit = one tone. The method of one
  bit = one tone was dropped for other forms of encoding when modems
  started supporting connection speeds in excess of 2400 baud. While
  the 8000 bit/sec sampling rate of POTS lines can theoretically
  support up to 4000 Hz bandwidth, it is technically impossible to
  get this (reliably) out of the analog phone system. 28.8K modems
  need at least 3200 Hz of bandwidth on a POTS line to support 28.8K.

  All said, I still stand by my comment that 53 modems per T1 is
  ludicrous.

  Dave Stoddard
  dgs@us.net

2

As long as we are being specific about this, I omitted data
compression too

V42bis compression is probably not very useful if the data is already
compressed either at the application level [gif, tiff, mpeg] or at PPP.

By the way, V42 [LAP-M] does not just add frame overhead. It is a full
blow error correction and flow control protocol and as such can add
sizable retransmission delay if the link is noisy and SREJ [selective
reject ... similar in concept to TCP SACK] is not used. Note that it is
a bit of an irony that even though an application decide to use UDP
because it specifically do not need or even want flow control and error
correction, componenet in the path such as the modem will go ahead and
do it anyway :frowning:

Regards,
John

3

For modem to modem communications, start and stop bits were used
up through 2400 baud when one bit = one tone. The method of one
bit = one tone was dropped for other forms of encoding when modems
started supporting connection speeds in excess of 2400 baud.

I assume you mean 2400 bps modems, i.e., v.22bis? Admittedly, it's been a
while since I used one, but my recollection is that 2400 bps modems operate
at 600 baud, with 4 bits/baud using QAM.

All said, I still stand by my comment that 53 modems per T1 is
ludicrous.

This, without a doubt, is true. :slight_smile:

4

This matches my recollection. I think it was 1200 bps that was the
last 1 bit per baud modem.

Cheers,
-- jra