Cisco CleanAir battles Wi-Fi smog
There's a long-standing need in wired networks to verify the quality of connectivity of wire, cable, and fiber with appropriate test equipment. But with wireless networks, this requirement extends beyond initial site surveys and occasional troubleshooting.
Managers of wireless networks have an ongoing need for spectral analysis, viewing not just protocol-based traffic but also energy in a given band. And this monitoring needs to be constant, because interference, can appear at any time.
At Farpoint Group, we refer to the analysis of energy at the physical layer to improve performance as spectral assurance. With the announcement of Cisco's CleanAir technology, which includes access points with spectral sensing as well as new management software features, spectral assurance is for the first time available as an integral part of the wireless LAN infrastructure.
We call this style of spectral analysis the centralized infrastructure model (CIM), differentiating it from the earlier but still-useful walking around model (WAM) typically implemented on notebook computers.
But instead of having to send out a trained engineer to detect, classify, localize and (perhaps) remedy a given interference challenge, all this can now be done continually and from a central location. So, in addition to getting round-the-clock coverage, the CIM strategy can help lower WLAN operating expense.
We tested CleanAir's functionality in four key areas:
1. Detection - identifying interferers and examining the duty cycle and intensity of their energy.
2. Classification - identifying a specific interferer by its electromagnetic signature. This enabled us to look for specific types of devices as we proceeded further through these steps.
3. Localization - via the power of a CIM-based installation to determine the location of the interferer, a new application of the location and tracking capabilities already present in many WLAN implementations.
4. Remediation - Eliminating the interferer, usually via automatically reconfiguring channel assignments. Localization can also be used to find and physically disable an interferer.
Our test configuration consisted of a Cisco 5500-series WLAN controller, three Aironet 3500-series access points with CleanAir Technology, an identical 3500 AP operating in "monitor mode,'' and a 3300-series Mobility Services Engine, which is an applications server used to perform location calculations.
We also used version 7.0.164 of Cisco's Wireless Control System (WCS) management software via a notebook PC. The 3500 AP is equipped with a Cisco-designed radio chipset that incorporates the sensor capabilities originally implemented in Cisco's notebook-based (WAM) Spectrum Expert product, and the version of WCS we used incorporates much of the software capability in Spectrum Expert as well.
Integrating spectral sensing and analysis directly into the infrastructure, however, allows automated remediation of interference situations, which is not possible with Spectrum Expert alone.
We used the following commercial products as interference sources:• Panasonic KX-TG3031S 2.4 GHz. cordless phone• GE 28021EF3-A 5.8 GHz. cordless phone• QSWLMCR 2.4 GHz. analog video camera• RC5833A 5.8 GHz. analog wireless camera• Nuts About Net's AirHORN signal generator, which puts out Wi-Fi-based interference on a pulsed, swept, or continuous basis.
We put one access point into what Cisco calls Monitor Mode, which allows it to scan all channels continuously, rather than also serving as an access point. A typical access point supporting CleanAir only monitors its own channel, and then only while receiving, so the added assurance of monitoring all of the local airspace can be quite valuable especially in detecting unauthorized devices.
WCS, however, does not provide any console functionality for Monitor Mode, which would be helpful, but its information is regardless applied, as appropriate, in remediation operations.
Finally, we entered the geometry of our test configuration into WCS. The three access points formed essentially an isosceles triangle with the long sides about 13.5 meters and the short side about 8.25 meters. This can be seen in the figure below along with both detection and localization information.
At power up, the system selects all channel assignments automatically, and we did not override this capability during testing.
We found the detection capabilities of the Aironet 3500 to be just as good (although not as feature-rich) as Spectrum Expert, which we used to verify performance during this part of the test. The 3500 was sensitive enough to pick up cordless phone and even microwave oven activity originating from nearby unrelated facilities, the closest of which was no less than 50 meters away.
CleanAir detected all of the above interferers, and all were properly localized via our simple geometric diagram. The system correctly identified each, and placed them on the diagram with reasonable accuracy. Some variability in location is to be expected given the vagaries of radio propagation, but we expect that network operations staff, guided by CleanAir's analysis, will have little problem finding identified interferers.
Each interferer was also properly classified, with the exception of the AirHORN. Cisco has not yet developed a classification model for this device, but the company reports that it is working on this. To be fair, the AirHORN is detected by Cisco's IPS feature.
Remediation in CleanAir is in the form of what Cisco calls event driven radio resource management (EDRRM), which, when enabled in WCS, automatically reassigns channels in response to detected interference. The algorithm for this appears quite sophisticated, as it prevents thrashing (rapid or continuous channel changes), and, indeed, channel changes occur only infrequently and may not occur at all in situations where such might be expected.
We found that the 2.4 GHz. video camera, which causes moderate to severe interference in this band for all channels except 1 and 11 (it is centered at channel 6), did motivate a change from channel 6 to channel 11. On the other hand, the 5.8 GHz. video camera did not cause interference sufficient to force a change of any of the 5 GHz. channel assignments.
There is little doubt that spectral analysis will become a fixture in the majority of enterprise-class WLAN installations over the next few years, as competitive products are introduced and refresh cycles proceed. The threat of disruptive interference - unintentional and otherwise - never goes away, and tools like Cisco's CleanAir clearly have the capabilities required to deal with the challenge.
We found that the product works as advertised, is easy to use with only minimal training and experience, and its rich feature set and very reasonable uptick in pricing of $300 over a non-CleanAir AP (the 1140) should generate significant interest from customers and competitors as well.
We expect that the range of devices characterized will increase over time, addressing our concern regarding the AirHORN, and that other conditions regarding the product line will be relaxed. For example, the MSE is required here only for location calculations; running these on a larger server also hosting WCS might address this cost and rack-space issue. Similar, remediation is today only supported when all access points are 3500s with CleanAir; at least partial remediation should be possible with any access point and some number of 3500s in monitor mode. We also previewed a capability in CleanAir called SE Connect Mode, which allows Spectrum Expert to use an Aironet 3500 as a sensor, enabling detailed remote analysis from any PC running Spectrum Expert. The client side of this functionality is due to released shortly.
Overall, we were quite impressed with the capabilities of CleanAir as implemented today, and look forward to further enhancements in this product line. Our initial testing here suggests that this capability is indeed valuable and belongs on the short list of features for any enterprise-class WLAN installation.
Mathias is a principal at Farpoint Group, a wireless advisory firm in Ashland, Mass. He can be reached at email@example.com.
Read more about anti-malware in Network World's Anti-malware section.