A baudy topic: Bits, baud, coding and politics

Steve recently made a shift from “residential DSL” to “business DSL” service, and that discussion will be followed up in another newsletter. But for today, we’re taking a step back into the history of telecommunications. Why? Because, being a typical “bleeding edge” adopter of new technology, Steve had to decommission his old DSL modem. And this happened to be a model that supported a user input of either Ethernet or ATM-25.

For the first part of the history lesson, there was a day – not all that long ago – when ATM was promoted both as a WAN service and as a LAN technology. Furthermore, in the day of 10Mbps Ethernet, 25Mbps ATM LANs were thought to be “fast.” But that raises the question: Why was 25Mbps one of the chosen “ideal” speeds?

Turns out it all has to do with bits, bauds, coding and politics. In common parlance, particularly for modems, “bits per second” and “baud” are often used interchangeably – and incorrectly. Technically, it’s incorrect to speak of a 56Kbps modem as a 56-K baud modem. Bits per second are exactly that. But, in a literal sense, “baud” refers to the number of changes in a line condition per second. If the only changes in the line condition are binary (on and off), then bits per second equal baud. But assume for a moment that a line can have four detectable line levels. In this case, one could have a bit rate that is four times the baud rate.

So what does this have to do with ATM-25? As it turns out, the proposal came primarily from IBM, which, at the time was using a certain chip set for 16Mbps Token Ring. And, the chip set used two baud for every bit, hence running 32Mbaud. As a technological advance, 25Mbps ATM was proposed to use the same chip set, but to use 4 out of 5 baud to represent bits. Thus, 80% of the “baud rate” was used for the “bit rate.” (By the way, the “real” bit rate for 25Mbps ATM was 25.6Mbps.)

We’ll leave the full discussion of why ATM-25 ultimately was not a great commercial success to another day. But a portion of the reason is that there was resistance to paying royalties to IBM to use its chip set.

We’re sure we’ll hear from you with other war stories on the “standards.” But suffice it to say that standards are based on more than optimal technical solutions.