[e2e] narrow vs. wide bandwidth (hz)

Cannara cannara at attglobal.net
Tue Oct 7 11:19:27 PDT 2003

Michael, your metaphor of "bandwidth consumption" isn't really all bad. 
Frequency allocations have indeed been thought of like land subdivision, as
David implies, and there's a finite amount of either.  Just as it's possible
to design an antenna to send multiple signals in the same band (e.g., via
polarization differences), it's possible to have multiple uses of frequency
bands, with particular bandwidths (I only use the Hz sense).  But, the key
here is again the signalling technique -- the symbology.  Once a particular
symbol set has been used in a band, that's it, until that use terminates. 
CDMA and other energy-spreading techniques indeed appear to give multiple
usages of the same band, but any given coding sequence exhausts that use
within whatever area the signals can be detected -- "detected" means by all
posible means, including error correction.  This is one reason why cell phone
connections aren't always possible.

At any given instant, any system can be using the narrowest bandwidth to
transmit a symbol, which is good, since others can then use another part of
the band.  Hopping around by a coding technique doesn't preclude being
economical of bandwidth when each symbol is transmitted, even if symbols are
sometimes lost to random conflicts and perhaps reconstructed by error
correction.  This is one reason why the original idea of Ultra-Wideband (UWB)
Harmonics from Hell has been objected to, because it simultaneously interferes
with many bands, even if only over very short ranges.  If we were to walk into
a server-clogged data warehouse that uses true UWB and our cellphone with GPS
fails to work, we wouldn't like that.  So being economical of bandwidth is
good, whether on a symbol-by-symbol basis, or the traditional, continuous-wave
of radio & TV.  This is the interaction of symbol choices and data rates --
how much bandwidth is needed to allow transmission of the symbols that best
work with the medium (noise, interference, etc.) and the data encodings
available to provide the desired data rate.


Michael B Greenwald wrote:
>    Mon, 06 Oct 2003 08:01:02 -0400
>    "David P. Reed" <dpreed at reed.com>
>    Michael - actually, metaphors are REALLY dangerous, and you are falling
>    into a trap here.
> Definitely possible.  I'll try to explain the nature of my
> [mis]understanding below.
>    At 06:41 PM 10/3/2003, Michael B Greenwald wrote:
>    >If you consider two systems that
>    >have identical properties in all respects except that one has narrow
>    >bandwidth than the other [I know this isn't always possible], I meant
>    >that you'd prefer the system with narrower bandwidth.
>    This is precisely the point I was saying is wrong.   You have fallen into
>    the trap of thinking that bandwidth is a resource that is consumed by a
>    communications system to produce its product.   Physically this is just not
>    true.  The metaphor persists in the minds of engineers because they have
>    been working for so long on FDM systems at the top level regulatory regime,
>    that they assume it must be true.
> OK; I would not say "consumed" (as in a manner that "exhausts" the
> resource, or uses it in some mutually exclusive way), but I *would*
> say "utilized".  And I would obviously only consider narrowing
> bandwidth(hz) relevant in systems where bandwidth (hz) is a limited
> resource (due to sharing, or any other reason).  In that context I
> assumed that if you have two implementations with all desirable
> properties equal, and equal in all respects other than bandwidth (hz),
> then the one that used narrower bandwidth (hz) would be better.  (And,
> of course, if using a wider band yields a better system, I will prefer
> the system that uses more bandwidth, but I thought I was clear that I
> was speaking about the [sometimes hypothetical or mythical] case of
> "all other things being equal".) If that's wrong, then I _have_ fallen
> into the trap you are concerned about, and here's where I'd welcome
> enlightenment (you can delete end2end-interest, but I may not be the
> only one who has fallen into this trap).  And I'm aware of spread
> spectrum, CDMA, etc., but perhaps I don't understand as well as I
> thought, or I may be misunderstanding something fundamental about your
> point.
> Awaiting enlightenment (or, at least, refinement of my terminology or
> improvement of my metaphors) ...

More information about the end2end-interest mailing list