Crash Course: WLAN Spectrum Analysis

Your wireless users are raving about their increased mobility and freedom with your now-operational WLAN, but celebration is the last thing on your mind: Your IT team has been stuck manually tracing random disconnections in the conference room--to a leaky microwave oven in the break room.

WLAN interference problems like this may be few and far between, but a solution isn't always apparent and the effects can be devastating. A more efficient but often-overlooked troubleshooting solution is a WLAN spectrum analyzer. And a new generation of user-friendly, mobile WLAN tools is making identifying WLAN interference easier.

Physical Challenges

A spectrum analyzer lets you monitor and troubleshoot the physical layer (radio waves) of your wireless network. It gives you visibility into Layer 1 (physical layer) of the network, much like a packet analyzer lets you view Layers 2 to 7.

In a perfect world, your WLAN operates much like your wired Ethernet LAN and requires only an initial install and verification to provide years of faithful service. Although the protocols and applications flowing over wireless and Ethernet are typically identical, the underlying physical layer is the true differentiator. Unlike Ethernet, where transmission takes place over a guarded and closed medium, wireless communications cross a vulnerable no-man's land where there are no guarantees: IEEE 802.11 networks operate in the same unlicensed spectrum as Bluetooth headsets, cordless phones and microwave ovens, and transmissions must contend with a slew of interference sources and physical obstructions before they reach their final destination. Just how these obstacles affect the WLAN varies from minor throughput degradation to, in the worst case, complete disconnection from the wireless network. (See "Running Interference", and for more on the common sources of interference, see "How to Block WLAN Interference" ).

Although many wireless networks operate sufficiently well for Web and e-mail applications, emerging high-throughput and real-time applications such as VoIP (voice over IP) on your 802.11 WLAN will magnify existing interference. 802.11 corrects for interference and packet corruption by retransmitting data, so any problems are relatively invisible to low-throughput and bursty applications like Web surfing. With more demanding services like VoIP and video, interference problems become more noticeable because these real-time applications depend on a steady stream of data. If a stream of VoIP packets is corrupted midtransmission, it can leave the caller with dead silence.

With the current overlay nature of wireless networks, users can overlook many of the physical layer problems: If they can't connect to the wireless network, they just plug into the wired network. But they will likely be less forgiving as wireless networks become the primary access medium and their connections get disrupted by interference.

ABCs of Radio Frequency

To understand spectrum analysis, you must first learn a few fundamental metrics used to describe any RF signal. The first is frequency, the actual location--measured in hertz--within the radio spectrum where the signal exists. You've likely tuned into a specific frequency, 95.1 MHz on an FM radio, for example, to pick up your favorite station. Wireless LAN devices do the same to communicate with an access point, except their operating frequencies are much higher, in the 2.4-GHz range for 802.11b/g and in the 5-GHz range for 802.11a.

Similar to frequency, the bandwidth of a signal is measured in hertz and describes how "wide" a signal is. Communication protocols such as Bluetooth are 1 MHz wide whereas 802.11b/g channels are 22 MHz wide--802.11b/g Channel 1, for example, is centered at 2.412 GHz but extends from 2.401 GHz to 2.423 GHz. It's important to know the bandwidth of a signal because it defines how much a signal overlaps with neighboring signals on adjacent frequencies.

Another element of RF is signal strength, or the relative "volume" of a signal. We could measure signal strength in a linear scale such as watts, but you'd end up comparing numbers like 1/1000 watt with other very small fractions of a watt. Most access points transmit at 100 milliwatts, for example, which is equivalent to 20 dBm (decibels in relation to 1 milliwatt). But by the time the signal reaches the client, the volume could be as small as 0.000000001 milliwatts, or the equivalent of -90 dBm. See how tough it would be to compare those numbers? By instead using the logarithmic scale of decibels, a larger range of numbers is represented in a meaningful manner. Wireless LAN signals most often are represented in dBm, with -30 dBm representing a very strong signal and -90 dBm representing a barely perceptible signal. The graph on page 75 shows a spectrum analyzer's view of a narrowband cordless phone signal.

Tools of the Trade

WLAN spectrum analyzers have evolved rapidly over the past year. I was using a standalone Avcom Ramsey analyzer about a year ago that required moderate RF experience with a cumbersome parallel port interface for exporting screenshots to a PC. Today I'm doing advanced spectrum analysis on my laptop with Cognio's Spectrum Expert for WiFi 2.0, previously the ISMS Mobile product. Cognio has shrunk nearly all the spectrum-analysis smarts into a single PCMCIA adapter, which is a more mobile and user friendly format than that of a conventional standalone analyzer.

But standalone spectrum analyzers, such as those from Anritsu and Avcom of Virginia, are still considered best by RF experts. These conventional analyzers typically offer a wider frequency range and a faster screen-update speed, but require a skilled operator who can recognize an interference signal from its waveform plot alone. Cognio has, however, made spectrum analysis more accessible to most IT professionals with a feature called interferer classification, a sort of antivirus scanner for your wireless network. By pattern-matching the frequency, bandwidth and signal strength of non-802.11 signals against its classification engine, the Cognio device can identify an interference source amid all the noise.

Although the Cognio tool makes spectrum analysis easier, it still comes with the conventional plots and charts that make expert spectrum analysis such a powerful skill. The real-time FFT (Fast Fourier Transform) plot, for instance, mimics a standard frequency versus power graph to provide precise spectrum information, and the sweep spectrograph charts this information over a period of time, color-coding the information based on signal strength. This is useful for tracking down a device that shifts in frequency.

Another useful plot is the FFT duty cycle, which analyzes how much time, in percent, a signal is being broadcast, or its duty cycle. An intermittent communication protocol like Bluetooth, for example, might have a duty cycle of 10 percent, whereas a constant transmitter like an X10 surveillance camera will have a duty cycle close to 100 percent. The higher a particular signal's duty cycle, the more time it spends transmitting, leaving little or no time for your WLAN to communicate on that frequency.

Although Cognio leads the pack in PC-based spectrum analysis, its Spectrum Expert for WiFi 2.0 comes with a hefty price tag: $3,995, about the same as a standalone spectrum analyzer. For smaller WLAN installations, MetaGeek offers a less expensive alternative with its $99 Wi-Spy, which includes many of the same plots as Cognio's product but isn't nearly as accurate, lacks interference classification and forgoes 5-GHz analysis. The Cognio product buys you added functionality you can't find anywhere else, and is especially useful for working with large WLANs and when reliability is imperative.

Finding the Culprit

The hardest thing about tracking down an interference-related disruption to your WLAN is the intermittent nature of most interfering devices. Users make or answer calls on their cordless phones or Bluetooth headsets with no discernable pattern, for example, and though a microwave, which operates in the 2.4-GHz band, is most often used at noon, it's never on all the time. A common best practice for spectrum analysis is to perform several spectrum sweeps at different times, which will give you a better chance of catching sporadic interferers. Luckily, most interference problems are located near the offending device, so you won't have to search the entire building.

The best way to pinpoint an interfering device is to divide a building into quadrants and take measurements in each one. From there, pick the quadrant with the strongest signal and divide it into four parts again, and perform a signal measurement in each sub-quadrant. Once you've narrowed it down that way, you can then perform a room-to-room search in the problem area, and more easily find the offending device, whether it's a microwave oven, cordless phone or other wireless device.

With Cognio's Spectrum Expert, the device finder tab graphs an interferer's signal strength over time. By walking around the building and playing a game of "hot and cold" (see the screenshot, at right), you can quickly find the offender.

Once you've identified the culprit, the easiest solution is to remove the interferer and replace it with something less offensive. If it's a 2.4-GHz phone, for example, replace it with a 900-MHz one, or if it's a Bluetooth headset, switch to a wired, hands-free headset. With medical or industrial equipment, sometimes the only option is to shield the device with metal. Another relatively simple alternative involves changing channels on your wireless infrastructure: If a leaky microwave oven disrupts communication on 802.11b/g Channel 6, try moving your equipment to Channel 1 or 11.

If an interference source is out of your jurisdiction, such as a cordless phone owned by the tenant next door, your only option is to switch your WLAN to a different channel. A trimode 802.11a/b/g wireless LAN can communicate in both 2.4 GHz and 5 GHz. Since a majority of interfering devices operate in 2.4 GHz, a 802.11a WLAN operating in 5 GHz will enjoy a less-crowded spectrum space and offer a wider array of nonoverlapping channels--presently there are 12--to choose from.

Distributed Future

Although show-stopping interference problems are rare in today's WLANs, they will become more common as enterprises move toward larger deployments, increased user load and more demanding applications. Mobile tools such as Cognio's laptop-based Spectrum Expert and Berkley Varitronics Systems' BumbleBee, based on Hewlett-Packard's iPAQ Pocket PC, are sufficient for finding today's problems, but these tools are typically deployed only after a WLAN disruption occurs.

A more proactive approach to preventing WLAN interference is to deploy an overlay distributed sensor system, such as AirMagnet's Enterprise 7.0, that lets you view the wireless spectrum remotely rather than conduct walkabouts. Although AirMagnet's spectrum analysis integration doesn't yet aggregate alarms to a central console, this feature probably isn't far off, further justifying the $2,000-per-sensor price tag. And eventually, access point, intrusion detection and spectrum analysis functionality will be integrated into one infrastructure, providing centralized monitoring and management across your entire WLAN.

Jameson Blandford is a lab associate at the Center for Emerging Network Technologies at Syracuse University.