I first wrote about RF Spectrum Management (RFSM) almost five years ago, and my feeling at the time was that, while very difficult, management of wireless LANs at the physical layer was not only possible, but would ultimately become a competitive battleground. Over the years, essentially every enterprise-class WLAN systems vendor developed their own take on setting channel assignments and transmit power levels, but really taming the airwaves remains challenging, partially because the bands used by WLANs are unlicensed and therefore subject to both interference and strict limitations on transmit power, and partially because the client-in-control protocols used  in .11 simply don't allow the degree of control one would expect from a fundamentally cellular architecture. Allowing clients to decide with which AP to associate and when to roam is madness. But that's the legacy of a standard developed much too early, before the usage models of WLANs had a chance to evolve. Ad-hoc was assumed to be the connection mode most users would apply, rather than today's world in which infrastructure mode is clearly dominant for obvious reasons.
And yet, clever engineering can work around almost any problem, and we're now seeing a new set of capabilities from Aruba Networks, called Adaptive Radio Management 2.0, that really do result in meaningful improvements in both control and throughput especially in dense (in terms of both APs and clients) wireless LAN environments. I had the opportunity recently to observe and authenticate some testing done by Aruba at the University of Washington. We saw meaningful improvements in throughput, as measured by the IxChariot benchmark, in one case by over 200%. And this was in a mixed-mode, mixed-client environment with no configuration of or modifications to clients of any form. We also observed great performance in the subjective testing of applications typically found in education settings.
Although Aruba wouldn't tell me how it works (it never hurts to ask, but the answer is quite often little more than a polite smile), it's pretty clear they are applying basic queuing and scheduling theory, balancing load among radio channels available on APs ("servers" from a theoretical perspective), and pushing high-demand traffic to the front of the queue - things we've been doing in operating systems and transaction-oriented applications for years. Given that this is wireless, though, there may be a bit of rocket science in there; Aruba mentioned patents to me. But it's great to see that the vendors are working nonetheless to eliminate bottlenecks throughout the WLAN value chain. 802.11n certainly helps here as well, but compatibility mode creates problems and, ultimately, we will run out of gas with brute-force solutions alone at the physical layer. Traffic scheduling and related techniques are thus likely the best answers when it comes to improving aggregate throughput, and putting solutions like ARM 2.0 in place now, before channels get clogged and network managers have to devote time and resources to figuring out what's wrong, makes a lot of sense. And while I expect new facilities in the standard to help in this area, most notably 802.11v (Wireless Network Management) and 802.11k (Radio Resource Measurement), I suspect, as is the case with 802.11s, that vendor-specific solutions will dominate regardless. Standards really are often the lowest common denominator, especially since the vendors creating those standards aren't going to give away the good stuff. They need it to compete, and, boy is the market ever competitive out there.
I wrote up our findings in a new Farpoint Group Technical Note, which you can find here. And I look forward to further advances in the concepts Aruba has productized over the next few years.
Mathias is a principal at Farpoint Group, a wireless advisory firm in Ashland, Mass.
The Leader in Network Knowledge
