Putting PHY-level analysis into the WLAN infrastructure
AirDefense sent me one of their 1U appliances and a few Model 520 sensors for possible inclusion in the Network World site survey test. While it didn’t make sense to include these products there, I did take advantage of the availability of the hardware to check out the new spectrum analysis feature in version 7.3.3-12, which was not shipping when we did the Network World spectrum analyzer test. Bottom line: this could be the start of something big.I guess everyone knows by now that I consider spectral analysis to be a key requirement for essentially every enterprise-class WLAN installation. This is the only way to debug PHY-related problems, and having a portable spectrum analyzer has saved my butt many times in diagnosing otherwise-invisible problems. Still, I’ve had on my wish list for some time the desire for spectral analysis (actually, I prefer the term spectral assurance in this context) to be part of the infrastructure, essentially in every sensor (or perhaps AP) in a given system. There’s a problem here, of course, actually several: cost, complexity, needing to scan all Wi-Fi channels, and integration. Is this part of the Wi-Fi system itself, or part of the IDS/SPS/assurance system, or something else altogether? Let’s not worry about that for the moment; let’s just agree that spectral assurance needs to be part of the infrastructure, and such a path will perhaps eventually replace handheld analyzers in larger installations.So it was exciting to get some hands-on time with AirDefense’s implementation. This is an optional feature, but it’s nicely integrated with other functionality and very easy to use. The display shows a spectrogram by frequency and time, and it’s possible to specify an “interference scan” with a longer dwell time on a smaller number of channels. Duty cycle is continuously updated. Threshold is adjustable to eliminate distant detected but likely non-interfering sources. I used the Nuts About Nets AIRHorn to generate interference and results were as expected. Scanning is a bit slower than many of the dedicated spectrum analysis products, but the result is still very usable.The only downside is that a given sensor can be, as expected, perhaps, in spectral or IDS mode at any given moment in time, but not both. It might make sense, then, to buy a few more sensors and dedicate them to spectral analysis. Regardless, I’m very encouraged, and it’s nice to seen the idea of infrastructure-based spectral analysis validated at last. On my wish list is this type of integrated spectral analysis with the ability to localize particular sources per locally-defined rules. RTLS thus once again becomes a central element in WLAN installations. More on this next week.
