[NCLUG] Wireless cards and Linux

[John L. Bass]

Wireless devices in some respects are plug-and-pray (barrowing the Plug-and-Play)
spoof. Signal propagation, interoperability, and co-location are almost magic
black box issues since the physical charactersitics can not be seen, heard or
felt. This leads to mis-application of experiences, and the propagation of those
experiences into other, but very different, environments where the experiences
no longer apply. It is very important when building wireless networks/topologies
to not generalize.

Take for example, the comment earlier about the relative ranges of two different
vendors equipment inside a building and down a street. Without a doubt, that
experience, at that time, with that equipment locations is valid. The problem
is that you don't know if it was strictly from signal propagation issues of the
systems under test, or if interference from other nearby co-located systems was
to blame.

The difficult problem with 2.4Ghz wireless, is that move everything 1/2 a wave
length (about 2.3 inches) and you can get completely different results due to signal
propagation issues. In free space (open air, no objects in or around the signal
path) the signal propagation is completely bounded by the phyics that the power
decays with the square of the distance. In cluttered space, all bets are off,
objects attentuate, reflect, focus, scatter, and refract the signal to to produce
a 3 dimensional nightmare of wildly varied signal intensities and qualities (AKA
multi-path). Add to this the power from nearby co-located systems, and the nightmare
is a lot worse.

Higher quality Radios us two antennas with wide mulitiple 1/2 wave length
spacing (diversity antennas), which in cluttered environments double the chances
that one of the two antennas will have a propagation pattern that is usable. Devices
with a single fixed antenna often require moving one or both ends around a few inches
to determine where the multiple wave fronts constructively add togather for a usable
signal (AT BOTH ENDS).

There is also the issue of transmitter power and reciever sensitivity. 30 milliwatt
radios have half the range (or less) of 100 mw radios. And due to a power per bit
issue, slower wireless lan radios have a longer range than faster radios.  All of
this depends on the quality of the firmware/software to determine when it's having
trouble, and choose a reasonable power and data rate for the connection. The general
feeling in the industry is that more power is bad, since it increases the range
that co-located systems will interfer with each other - and create a poor image
for the consumer. Likewise with higher gain antennas, they push your signal farther,
but also allow interference from systems farther away.

Lastly external antennas can be a great aid in getting things to work ... since you
can often move the antenna away from the metal case of the server to get an even
signal distribution with minium reflected waves. Lucent offers the "range extender"
which does very well, and Aironet/Cisco a variety of antennas including directional
panels and diversity "rabit ears" for mobile notebook operations. The gotcha here
is that the FCC has very strict guidelines on antennas, since they are part of the
certification process required for the "radio system". Hacking an AirPort, or other
access point or radio to use antennas not provided by the manufacturer (or third
party manufacturer who has properly certified the radio-cable-antenna system)
is subject to $10,000 per day up to $75,000 (per device) plus a $100,000 personal,
or $200,000 business criminal penalty in the US.

So if you purchase 3rd party devices to build a mix-n-match configuration make sure
they have the proper FCC certifications for the particular radio system, or be aware
that you are looking at $175,000 or more in fines, if somebody complains and turns
you in. There are few exceptions to this process, all are for fixed installations,
and require professional installation to make sure the FCC regs are fully met. This
is why all the radios that offer external antennas have funny hard to get connectors,
as required by the FCC ... see below.

Inside CWX we use kitted systems from Wincomm with the required certifications,
or a combination of kitted systems with professional installation. Our members
can buy a fully kitted system with pre-made cables and do their own install to
meet this requirement. We do not allow adhoc mix and match/match installations to
protect both CWX and the member. This is because the FCC in other areas has
shutdown, and fined, similar wireless ISP operations that violated the FCC reqs.

John Bass

-------------- Extracted from current FCC Regs, Part 15  ---------------------

PART 15 - RADIO FREQUENCY DEVICES
Subpart C - Intentional Radiators Section 
Section 15.203 Antenna requirement.

An intentional radiator shall be designed to ensure that no antenna other than
that furnished by the responsible party shall be used with the device.  The use
of a permanently attached antenna or of an antenna that uses a unique coupling
to the intentional radiator shall be considered sufficient to comply with the
provisions of this Section. The manufacturer may design the unit so that a broken
antenna can be replaced by the user, but the use of a standard antenna jack or
electrical connector is prohibited. [...] Further, this requirement does not apply
to intentional radiators that must be professionally installed, such as perimeter
protection systems and some field disturbance sensors, or to other intentional
radiators which, in accordance with Section 15.31(d), must be measured at the
installation site. However, the installer shall be responsible for ensuring that
the proper antenna is employed so that the limits in this Part are not exceeded.

Section 15.204 External radio frequency power amplifiers and antenna modifications.

(c) Only the antenna with which an intentional radiator is authorized may be used
with the intentional radiator.