[PLUG] homegrown WAP

[Keith Lofstrom]

Carla, if you are planning to do a bunch of sites, it would be handy
to learn a bunch about WiFi radio propagation.  To make any money at
this, you will have to get pretty good at looking at the insides of a
building and eyeball estimating where the dead spots will be, or you
will have massive overruns in material and time.

802.11b operates at the same frequencies as microwave ovens, so if a
chunk of material gets hot in your microwave it will likely soak up a
lot of signal in walls.  Others have pointed out how metal or concrete
can block WiFi signals, which is why the back-of-PC antennas are
risky (Maxwell trumps Babbage every time).   Multipath means bouncing
signals will interfere with each other, leading to dead spots in the
damndest places.  That means more WAPs to fill in the holes.

Cables eat signals at these frequencies; that is why the professionals
mount the antennas in the best spot, move the WAP to the antenna, and
use long ethernet cables to move the lower-frequency, higher-signal-
strength data to the WAP.  Antennas with gain are touted as the fix by
some folks, but all "antenna gain" really means is that less signal
goes some places and more others.   If you want a cloud, you need
an omni with low gain.  If you want a sector, you can use a sector
antenna and put the cloud where you need it.  A tall antenna with 
high gain flattens the cloud;  you will get less propagation above
and below the plane.  A big parabolic dish gives super gain and a
very narrow beam.  Etc.  You are filling a volume with energy;  
you can shape it, but you can't expand it with an antenna.  Energy
is conserved;  Maxwell rules.

Whatever goes for 802.11b at 2.4GHz goes in spades for 802.11g at
5.0GHz, and the new 802.16a metro wireless.  Something that just
barely works for 11b is probably going to fail miserably for 11g.
Cable losses go up literally exponentially.  Since you want to
eventually peddle upgrades, it is best to do it right the first time.

A good book with some chapters on antennas and propagation is 
"Drive By WiFi Guide" by Jeff Duntemann.  There is an OK book with
more of the software aspects, including a chapter on Linux,
"The Book of WiFi" by John Ross.   And of course, the fine people at
Portland Personal Telco know all sorts of useful things, and much 
good info can be found at http://www.personaltelco.net .  Invictus
Networks probably peddles all the gear you need;  good folks with
reasonable prices.  

The "Kismet" program is your friend - it can give you a crude measure
of signal strength.  A high power WAP is not much use if it does not
have good receive sensitivity - the link is bidirectional, and you
need good signals both ways.

Keep in mind simple electricity cost.  Even if you don't give a rat's
hind end for salmon, a PC consuming 150 watts is using 1300KWhr/year;
that's about $100/year for electricity.  Reducing that to 10-30 watts
can save a bunch of power and money, and leave a little more water
behind the dam to float a fish or two.  That's why an old laptop (with
the screen turned off or removed) for an AP is a good idea, a Soekris
box is a better idea, and an off-the-shelf WAP not a horribly bad idea,
especially if you can run Linux on it.  Old PCs containing slow, power
hungry, and marginally reliable hardware,  anchored to a floor,  are
not a good idea at all.

Good luck and happy hunting.  There is both science and art to this
stuff;  you can't make a profit at it until you know both.

Keith

-- 
Keith Lofstrom           keithl at ieee.org         Voice (503)-520-1993
KLIC --- Keith Lofstrom Integrated Circuits --- "Your Ideas in Silicon"
Design Contracting in Bipolar and CMOS - Analog, Digital, and Scan ICs