Aruba Mobility Controller Upgrade Tames Unruly Wi-Fi Clients

By Lisa Phifer

September 30, 2008

Adaptive Radio Management (ARM) 2.0 uses standards-based infrastructure controls to force self-centered Wi-Fi clients to share the air more efficiently.

Let's face it: Wi-Fi clients are greedy. Designed to think only of themselves, 802.11 stations associate to the best available AP and send data as fast and as often as they can.

In lightly-populated WLANs, nobody notices because everyone is happy. But in dense or over-subscribed deployments, selfish clients can hog bandwidth, overload APs, and drag down total throughput. Adaptive Radio Management (ARM) 2.0, a free upgrade for Aruba Mobile Controllers, uses standards-based controls to tame unruly clients for the common good.

"Clients have a one-dimensional view; they lack the system-level perspective needed to optimize overall WLAN performance," said Mike Tennefoss, Head of Strategic Marketing at Aruba Networks. "ARM 2.0 uses infrastructure-based controls and existing standards to tame client behavior, forcing them to do what is in the best interest of the entire network."

Hands-off adaptation

In ARM 1.0, Aruba did what many other vendors have done: capitalize on a WLAN controller's perspective to optimize AP behavior by auto-adapting channel and power settings. In ARM 2.0, Aruba focused on auto-adapting client behavior to optimize aggregate performance.

Aruba is not the first to tackle this challenge. Some of the earliest "fat APs" used proprietary load balancing protocols to re-distribute clients. Hooks that redirect stations inside homogenous WLANs can also be found in WLAN client software (e.g., Cisco CCX AP-assisted roaming).

But, unlike those approaches, ARM 2.0 requires neither client software nor protocol extensions. Instead, ARM 2.0 uses standard 802.11 management and control frames to steer clients onto desired bands/channels/APs and influence PHY data rates. ARM 2.0's "secret sauce" is Aruba's patent-pending process that distills client capabilities and makes decisions to improve network throughput without unduly penalizing most individuals.

For the common good

The infrastructure-based controls implemented by ARM 2.0 include the following:

·     Band steering: Helps dual-band 802.11a/n devices experience up to 200% better throughput by discouraging their use of the 2.4 GHz "junk band."

·     Spectrum load balancing: Stops a single AP from being hammered by rejecting selected clients, thereby encouraging association to another less-utilized AP.

·     Coordinated channel access: Discourages stations that detect collisions from dropping their PHY data rate when doing so would not actually benefit them.

·     Co-channel interference mitigation: Converts under-utilized APs into passive monitors to reduce interference when capacity exceeds demand.

·     Airtime fairness: Allocates more AP time to servicing faster clients (n > a/g > b) so that older clients cannot dominate the air at their expense.

·     Performance protection: Reduces the adverse impact of 802.11b protection mechanisms on newer a/g/n clients.

Some controls can be selectively enabled—for example, you might not want to steer clients onto the 5 GHz band. But otherwise, ARM 2.0 is a hands-free adaptive radio management system. Customers can simply upgrade Aruba Mobility Controllers with free ARM-capable software (available 4Q08). Once ARM 2.0 has been turned on, no manual tuning is required.

Road-tested

To demonstrate ARM 2.0, Aruba engaged the Farpoint Group to conduct a series of tests at the University of Washington. These in-situ tests were not formal benchmarks, obtained in a controlled test lab. Rather, they illustrated how one densely-populated real-world WLAN could benefit from infrastructure-based controls.

Tests were conducted in a large lecture hall, half-filled by 100 wireless laptops. Clients included a diverse mix of vendors, chipsets, operating systems, and capabilities, ranging from PC to Macbook, XP to Vista, Broadcom to Intel. Traffic was generated by several education applications, including DyKnow client monitoring, Video Furnace IP video streaming, ExamSoft on-line testing, and Classroom Presenter, a UW-developed electronic whiteboard.

Throughputs observed with and without ARM 2.0 showed clear impacts. For example:

·     When ARM 2.0 was enabled at 2.4 GHz only, a mix of 11b, g, and n clients generated 20 percent more aggregate throughput. 11g clients experienced a 65 percent boost, while 11n clients saw 20 percent improvement—at the expense of 11b clients, who did receive less airtime.

·     However, when ARM 2.0 was permitted to steer 802.11n clients onto the 5 GHz band, everyone benefitted. 11n throughput grew 217 percent; 11g by 162 percent; and 11b by 108%—an aggregate increase of 70 percent.

·     When 25 802.11g stations were tested without ARM, some bursty clients managed to send 2.73 Mbps more than others—some of whom were practically starved for attention. With ARM 2.0 enabled, standard deviation between client throughputs fell 30 percent.

·     When all 100 clients were tested simultaneously, ARM 2.0 divvied the load by placing between 14 and 20 clients on each of six available channels, supported by four APs. Even in this dense, crowded environment, all 100 clients managed to satisfy tough application demands (e.g., < 2 second slide refresh).

These are just a few examples; your own mileage will vary. However, these test results do appear to validate Aruba's ARM 2.0 approach to optimizing overall WLAN performance in this scenario.

“Wi-Fi transmissions are subject to a wide variety of impairments in the real world, and appropriate infrastructure control can improve client and network behavior with clear performance benefits,” said Farpoint Group principal Craig Mathias. “As we saw in testing at the University of Washington...the benefits of ARM 2.0 are impressive.”

Lisa Phifer owns Core Competence, a consulting firm focused on business use of emerging network and security technologies. She has been involved in the design, implementation, assessment, and testing of NetSec products and services for over 25 years.

Originally published on .

Comment and Contribute
(Maximum characters: 1200). You have
characters left.