Telebit Trailblazer modems (was Re: Debugging techniques and core dumps)

From: Eric Smith <eric_at_brouhaha.com>
Date: Mon Feb 14 19:41:02 2005

Tom Jennings wrote:
> Great things those Trailblazers were! six THOUSAND! bits/sec to
> French Antilles or somesuch place. It was amazing.(*)

I wasn't one of the modem engineers, but those guys did amazing things.
Our modems could maintain connections to places where all other modems
failed due to phone lines with very poor frequency response and phase
distortion characteristics. Customers told us that they could
communicate with remote parts of Mexico, Central America, and South
America, where no other modems would work.

The Trailblazer did this through the use of a patented "PEP"
(Packetized Ensemble Protocol) multicarrier modulation, with 511 narrow
band carriers each at a low bit rate. IIRC each carrier operated at
approximately 6 baud (symbols per second), with a constellation of up to
64 symbols (6 bits per symbol), for a maximum bit rate of about 18 Kbps.
(Higher effective bit rates were possible through use of compression.)

Carriers with poor SNR or high phase distortion would train using fewer
constellation points (4 or 16), or not be used at all.

The major drawback to PEP was that it was half-duplex. There was
a low-rate back channel, but if you needed to send much data in
the reverse direction, the modems had to negotiate line turnarounds.
This was done transparently, and the modems used handshaking with
the hosts to avoid data loss.

The Trailblazer became "the modem to have" for UUCP due to the high
transfer rate, and a special "UUCP G-protocol spoofing" feature.
Without spoofing, the UUCP G-protocol acknowledgement packets were
large enough to cause line turnarounds, which would kill the
performance. But because the modems implemented error control,
the feature was added to recognize G protocol, and spoof the ACKs,
avoiding the turnarounds.

Later the modem group did some research into full-duplex PEP, but
it never made it into a product. By that time, the majority of
the modem market didn't want anything but V.32 modulation.

PEP was the forerunner of DMT modulation used by most DSL modems,
and OFDM modulation now used for many high-bandwidth wireless
applications including 802.11g. Telebit believed that DMT required
the use of the PEP patent, and that they would be able to collect
a lot of royalties from licensing, but AFAIK that never happened.
If they had pushed it, and if the PEP claims were found to apply
to DMT, the carriers probably would have just used CAP modulation
instead, as some did anyhow.

The Telebit modem engineers believed that it would be possible to get a
working connection using a tin can and string (with some sort of
transducers, perhaps piezo), but never actually did the experiment.

Eric
Received on Mon Feb 14 2005 - 19:41:02 GMT