[e2e] Bandwidth Estimation workshop

[Enrique Hernandez-Valencia]

Another data point,

Part of the confusion on the usage of the terms bandwidth, baud rate,
bid rate etc., is context specific.

For instance, traditional telecom theory has had the concept of
bandwidth to refer to the frequency consumed by a particular "signal",
or, in the information theoretical sense, symbol. The relationship
between freq (in Hz) and symbol rate being modulation technique-specific
(AM, FM, PSK, QAM, etc.)

Thus, for each particular modulation  technique, there is
a 1-to-1 mapping between the bandwidth in hz and the symbol rate
(used to be referred to as the baud rate).

When the symbol space is drawn from a binary space, each symbol
represents a bit of information and baud rate = bit rate. If the
symbol space is size M then, bit rate = log2(M).

Thus, for a given technology, there is "typically" a 1:1 map between
bandwidth (hz) -> baud rate (symbol/sec) -> bit rate (info units/sec)

Bandwidth and bits are tangible to most of us. Baud, the
"glue" has been lost in the translation ...

Enrique



On 10/3/2003 11:35 AM, David P. Reed wrote:
> I proposed frequespan to replace the frequency meaning of bandwidth, 
> knowing that it was an ugly word and someone would probably come up with a 
> better one.
> 
> I wouldn't be unhappy with these:
> 
> 	bandspread (since "spread spectrum" and "spreading" are in use commonly)
> 
> and
> 
> 	bitrate (though "information rate" and "signalling rate" would break down 
> the ambiguity here between bits as information and bits as symbols).
> 
> Units for the
> 
> Of course, cultural change could take DECADES.
> 
>