The Wireless Edge: 802.11n for the Enterprise--Not a Trivial Upgrade

Six hundred megabits per second! That is the highest theoretical speed of IEEE 802.11n if a product implements all the mandatory, as well as optional, capabilities of the standard. The same core radio technology that enables these high rates also extends range, which results in increased AP (access point) coverage. Although speed and range are generally good things, there are ramifications, along with various other things to consider, with this standard. I bring this topic up now for two reasons. First, IEEE is planning to release the second draft (draft 2.0) of the IEEE 802.11n standard this quarter, with compliant products available shortly thereafter. Second, I'm teaching a course on Wi-Fi technology evolution and integration at Portland State University on January 26 (see http://www.rysavy.com/training.html for more details), so I've been busily creating and updating 802.11n slideware.

While this is a good time to start understanding migration to 802.11n, in most organizations it's probably best not to start deploying until the ratified version of the standard is available. And it's not expected until 2008. But you can expect tremendous promotion of draft 2.0 products this year, particularly because the Wi-Fi Alliance will be certifying products for interoperability. I personally find the concept of certifying against a draft standard a little strange, but the argument is that the draft standard should be reasonably close to the final standard, that it already incorporates a tremendous amount of scrutiny and review, and that the draft standard offers powerful capabilities. Thus, the certifications will help develop the market faster as well as offer customers benefits sooner. For instance, certified cards will interoperate with certified access points. So, certainly for smaller deployments like consumer and SOHO, the draft 2.0 products will make sense. Although it is hoped that products will be either firmware upgradeable to the final standard or interoperable with final-standard products, there is no guarantee. In my view, deploying draft 2.0 in larger networks entails considerable risk, especially given the tortuous history of this standard.

Regardless of when you deploy, it's important to understand the speed and range capabilities so you can start planning now. I quoted 600 Mbps as the highest theoretical rate mostly to get your attention. This is indeed possible with 4 X 4 MIMO (Multiple Input Multiple Output with 4 transmitters and 4 receivers), a 40-MHz channel, a short clear view of the AP and some optional features implemented. But it is only barely theoretically possible. The actual mandatory requirement is 2 X 2 MIMO, which delivers a highest rate of 300 Mbps if all other features are implemented. So in the real world, as an educated guess, 2 X 2 MIMO throughput rates should turn out to be 100 Mbps to 200 Mbps, though actual rates across different types of environments remain to be seen. Still, we're talking a massive increase in speed compared to 802.11a or 802.11g, which max out at 25 Mbps of effective throughput.

You should think about how much speed you actually need. If you're surfing the Web, you'll likely be constrained by your Internet connection. But for sending video streams around homes, the bandwidth will be extremely useful. In work environments, higher throughput translates to more users per access point and better operation of services like VoIP, though these need a variety of additional enhancements such as QoS and fast roaming to be effective.

Where 802.11n speed impacts planning is with your Wi-Fi infrastructure. First, any decent 802.11n AP should have a Gigabit Ethernet connection. This may seem obvious, but some early products only supported a 100 Mbps connection. Next, if you're using a Wi-Fi controller architecture, can the controller support APs operating at much higher throughput rates? There may be backplane or processing limitations. Interestingly, some vendors--Trapeze Networks, for example--are moving some functions, such as cryptographic processing, back to the edge. This reverses the recent trend in the industry to centralize as much functionality as possible. At 802.11n speeds, it doesn't necessarily make sense for every packet to have to flow through the controller. So you'll need to see what precise migration plans your vendors may have for supporting 802.11n. While you're at it, ask them if access points can be upgraded to 802.11n by replacing radio modules or whether you'll need entirely new APs.