Cisco Cisco Aironet 3500p Access Point Weißbuch
© 2014 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public Information.
Page 5 of 11
Once SAgE processing has completed, the radio samples of interesting pulses are passed to the software level for
detailed fingerprint analysis. Performing this processing on the main radio CPU would negatively affect Wi-Fi
performance. To eliminate this impact, the Cisco hardware solution includes a custom processing core called the
Digital Signal Process
or (DSP) Vector Accelerator (DAvE), which is integrated directly into the access point’s Wi-Fi
chipset. The DAvE core is able to perform intensive signal processing operations, referred to as “Davelets” - such
as filtering, decimation, rotation, sync-word detection, and modulation detection - without burdening the main CPU.
The DAvE handles CPU-intensive signal processing operations that would otherwise be a burden to the main CPU.
The final processing level occurs in a software module that runs on the main
CPU and is referred to as “Sensord.”
Note that because the heavy lifting has been accomplished by the SAgE and DAvE hardware blocks, the overhead
on the CPU is now very low. The Sensord software looks at the timing and frequency of interference bursts, and
the discovered attributes of the bursts such as the modulation type and identified sync words. This high-level
information is then used to perform the final identification and separation of one device from another. This final
classification step provides the powerful features of SI: telling you the specific source of the interference, where it is
located, and how it can be mitigated.
Performance Aspects of SI Implementations
Number of Classifiers
Cisco CleanAir technology includes a robust suite of 20 non-Wi-Fi classifiers. Because the analysis takes place in
software, the list of classifiers may be expanded as new interference sources become relevant in the market.
In other words, the solution is capable of detecting any kind of interference that might be introduced in the future,
and requires only a software update.
Simultaneous Detection
Cisco CleanAir technology classification is able to distinguish several different interferers - either of the same type
or different types - that are operating at the same time. In fact, CleanAir technology is capable of tracking and
classifying hundreds of devices and reporting the top 10 simultaneous interference devices per radio. This is
important, because in the real world the amount of simultaneous RF activity can be quite high. Any competing
solutions that are not sophisticated enough to distinguish multiple simultaneous devices will quickly fall apart in the
field and are only good enough for demos and lab tests.
Time to Detect
Interference devices can be transient, either because they are turned on and off quickly, or because the user is
moving through the floor space. For this reason, classification must occur quickly, before it is missed. Cisco
CleanAir technology enables access points to classify devices within 30 seconds, and is often able to perform
classification in less than 5 seconds. (Note that reporting may be slightly delayed when consolidating data across
multiple access points).
Probability of False Detection
It is important to not miss a source of inte
rference, and it’s equally important to not report “phantom” interference
when none exists or to mislabel interference, causing IT to look for the wrong type of device. Cisco CleanAir
technology is designed to produce low false-detection rates, even in very busy RF environments where hundreds
of Wi-Fi and non-Wi-Fi devices are operating simultaneously. By reducing false detection, CleanAir technology
saves IT time.