Nothing frustrates the logical minds of network engineers and their
kindred than dealing with the vagaries of wireless. A dark art it is,
as they tussle with the intangibles of multipath, interference and
fluctuating performance.
It's not that there aren't mature tools available to help apply the
scientific method. AirMagnet's Laptop Analyzer, for instance, provides
good layer 2 analysis tools that slice and dice channel usage, frame
types, list 802.11 devices and their associations, and more. WildPacket's
AiroPeek NX, which focuses a bit more on layers 3 and up, provides
histograms of packet size and graphs of top talkers as well as good
old packet captures, decodes and expert analysis. And most
enterprise-class wireless infrastructure solutions can generate heat
maps that show coverage with predictive link rates and SNR (signal to
noise ratio) levels as measured by the access points.
The challenge with Wi-Fi, in both the 2.4- and 5-GHz bands, is that it uses
unlicensed spectrum that must not only compete with other devices in
the same frequency band but also accept any generated interference.
There's a long list of 2.4-GHz interferers: microwave ovens, cordless
phones, Bluetooth devices, baby monitors, wireless video cameras, certain
cordless mice, wireless SCADA (supervisory control and data acquisition)
equipment, wireless mesh backhaul and, of course, neighboring access
points. Some of these items are idle most of the time (microwave ovens),
while others may always be active (wireless video cameras). Whatever
the case, these competing devices can use some or all of the spectrum
in the unlicensed band and generate interference that reduces performance
by raising the noise floor, which in turn reduces the coverage. Competing
devices can also inadvertently affect a DoS (denial of service) attack
because either the Wi-Fi device can't pass the requisite "clear channel
check" before transmission or most received frames are corrupted.
SOHO (small office, home office) wireless vendors such as Rotani and
Ruckus claim their Wi-Fi products work around that, but these products
have not entered the enterprise space nor do they locate or identify
the source of the interference.
Network engineers who want a view of the physical layer (which in
wireless is electromagnetic) have had to spend anywhere from $5 thousand
to $40 thousand for traditional spectrum analyzers. These heavy, boxy
units with smallish screens plot spectral energy over an adjustable range of
supported frequencies. For example, you can look over the whole 83.5 MHz
used by 802.11b/g devices or zoom in on just one 22-MHz-wide channel.
And it's not as if these units do not sport serial output interfaces, include
local data storage or offer input connectors. Rather, they are bulky
instruments that perform no expert analysis.
Enter former chipset company Cognio, whose first chipset work focused
on advanced wireless technologies such as beamforming (although market
externalities led the company to spin off that technology). Related work
on interference detection led to Cognio's development of a spectrum
analysis chip that forms the basis for the company's current lineup of
mobile RFID, mobile Wi-Fi and distributed Wi-Fi spectrum analyzers.
Spectrum Expert for Wi-Fi, the company's standalone product, consists
of a Cardbus-based PCMCIA card and related software that is able to
examine swaths of spectrum up to 200 MHz wide (unlike traditional
access points that cycle through channels one by one) and track as
well as identify sources of interference. For example, the basestation
of my 5.8-GHz cordless phone was identified in less than one minute
and a short list of matching models, of which my unit was one, followed.
This kind of identification is not possible with Wi-Fi devices or traditional
spectrum analyzers. And, as my colleagues found in a formal evaluation
of this product,
Spectrum Expert for Wi-Fi can identify and locate devices you didn't
even know existed.
