Xref: utzoo comp.dcom.modems:2640 comp.mail.uucp:2128
Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!rutgers!uwvax!oddjob!mimsy!chris
From: chris@mimsy.UUCP (Chris Torek)
Newsgroups: comp.dcom.modems,comp.mail.uucp
Subject: Re: UUCP g stats / baud isnt bps
Message-ID: <13924@mimsy.UUCP>
Date: 9 Oct 88 18:55:41 GMT
References: <183@arnold.UUCP> <1988Sep20.184054.2403@utzoo.uucp> <184@arnold.UUCP> <210@obie.UUCP>
Organization: U of Maryland, Dept. of Computer Science, Coll. Pk., MD 20742
Lines: 51

>>In article <1464@netmbx.UUCP> blume@netmbx.UUCP (Heiko Blume) wrote:
>>>as far as i remember the baud rate tells you how many changes of the
>>>discrete signal levels are possible.

>In article <13842@mimsy.UUCP> I answered:
>>Not exactly.  `Baud rate' means `number of channel symbols per unit
>>time'; multiplying this by `number of bits per channel system' gives
>>the more familiar bits per unit time, typically bits per second.

In article <210@obie.UUCP> wes@obie.UUCP (Barnacle Wes) replies:
>Nope, Heiko was right.  Baud refers to the number of state changes the
>carrier makes per unit time.  How this relates to the bit rate depends
>on the modulation scheme you are using (NRZ, NRZ1, diphase, etc.).

All three of these definitions are different.  While it is not my
field, I must say that, from an information-theoretic point of view, my
definition makes the most sense to me (no surprise here :-) ).
Heiko's definition is `how many
changes are possible'; your is `how many are made per time'.  It is
easy to convert from `possible changes *per* *time*'---which is no
doubt what Heiko *meant*---into `amount of information per time', but
knowing the actual number of changes (your definition, unless you
reword it) tells very little.  Imagine a 300 baud FSK modem that is in
the spacing state, and remains that way for .02667 seconds.  It is
making *no* state changes per unit time, yet a machine connected to it
infers from this a NUL character---eight bits of information!

With my definition, this 300 baud modem is sending 300 channel symbols
per second.  Each channel symbol is one bit of information.  A 1200
`baud' 212A modem sends 600 channel symbols per second, but, since it
uses quadrature modulation, carries two bits of information per symbol,
for a total of 1200 bits per second.  Telebit's DAMQAM sends very few
channel symbols per second, but each one carries many bits.

There appear to be three orthogonal issues:  Imprimus, the number of
channels; secundus, the number of channel symbols per unit time per
channel; and tertius, the amount of information carried per channel
symbol per channel.  This is why I like my definition: it fits the
situation.  (There is, of course, a bias built into this situation!)

>...  In other words, if you send out an ASCII NUL character, you
>will only get two state changes in the entire byte - one for the start
>bit, and one for the stop bit.  If you are transmitting at 19,200 bps,
>this give you an effective baud rate of 3840!

Well, that is certainly consistent with your definition of `baud'.
Anyone have an official reference handy, so we can find out who is right
(and if it is Wes, why the---to me---peculiar definition of `baud')?
-- 
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7163)
Domain:	chris@mimsy.umd.edu	Path:	uunet!mimsy!chris