802.11n is the most significant change since the 802.11 standard emerged, with across the board changes, including the physical and MAC layers, modulation and antennas. Enabling the standard is an Herculean task, with proposals from the volunteer group meetings eliciting thousands of comments, each resolved either by a written response declining it or text explaining its inclusion.
Along with the logistics, there are the usual politics and manoeuvrings, but the Alliance began certifying draft 11n products in summer 2008, lessening the risk for developers. 11n offers an acceptable level of interoperability between 11n devices – plus backwards compatibility with 11a/b/g, so a smooth migration to 11n should be possible. The date for 802.11n ratification is September 2009, promising a dramatic leap forward in capability: up to 300 Mbps raw data rate, up to four data streams, improvements in modulation from 54Mbps to 64 Mbps at 20 MHz, plus channel bonding (taking two defined channels and combining them, doubling the data rate to 40 Mbps).
Phil Belanger is a veteran of wireless broadband and known in much of the media as “Mr Wi-Fi”. A founder and one-time chairman of the Wi-Fi Alliance, a hands-on mover and shaker in the industry for over 25 years and now marketing and business development chief at “channel blanket” proponent Extricom. He is also a pragmatist: “As an industry, we risk over-hyping the capability and focusing too much purely on speed improvements. In reality, there is a range of speeds possible and I am concerned that the market will view that as disappointing at first sight.
“It is tempting to focus on the Mbps, but 802.11n is a complex animal. It can deliver improved range and coverage, but it is impossible to give a definitive answer for all circumstances: much still depends on the distance from the AccessPoint [AP].”
What is impressive however – he describes it as “fabulous” – is the backwards compatibility as a requirement of the standard: “Today 11n is pushed most strongly in consumer environments, which is where the early success are. High-end laptops now have 11n built in. 11n is definitely a positive experience from the consumer perspective. I put an 11n router in my house and found it worked at a similar speed to my legacy 11a/b/g products but at a far greater distance; an 11n client running on a legacy infrastructure will perform as well as the best a/b/g client, too.”
Give us the tools
Enterprises are, understandably, proving slower to adopt 11n – apart from education and, to a lesser extent, healthcare. There are important planning and control issues to consider – even wired network challenges because 11n amounts to a system at the edge of the wired network delivering several hundred Mbps and there is no uniform 11n client with a single performance profile: “It works beautifully, but it does mean we have to change how we plan and build enterprise networks, otherwise there is the possibility of bottlenecks elsewhere in the wired network and overloads.”
Using 11n’s MIMO (multiple input, multiple output) feature, the coverage pattern of 11n APs is unpredictable as each AP’s coverage area is erratic due to the way 11n works; if you fill in all the coverage “holes” by turning up the power, it attracts co-channel interference. Belanger highlights the need for new tools for tuning an design to meet the challenge of 11n: “The AP coverage pattern will be different for each client type, making AP cell planning as much an art as a science. Equipment providers and integrators are just beginning to learn best practice for 802.11n”
As to the choice between 2.4 GHz and 5 GHz, Belanger points out that most enterprise APs use 11n radios that can tune to either band, although not dynamically, so it is an important choice. The Extricom system in the 2.4 GHz band (with three non-overlapping channels), for example, could use 2 x 40 Mbps and 1 x 20 Mbps for legacy; at 5 GHz there are more channels available but more variables as to performance, such as walls and distance.
Security blankets
Extricom favours its channel blanket (rather than a micro-cellular) AP architecture, controlling the APs via a centralised switch. The approach advocates configuring 11n’s three non-overlapping channels at 2.4 GHz to offer three independent “blankets” operating independently, so one could be exclusive to public internet access, for example, or for encrypted data on an entirely separate VLAN and WLAN.
Belanger says that while is not part of the 802.11n standard, the approach is complementary with 11n’s spiky pattern AP coverage challenge, with adjacent APs on the same channel, lots of overlap and the central switch eliminating co-channel interference. The APs are at the physical layer while the MAC layer is controlled by the central switch, coordinating the actions of the APs and not allowing adjacent APs to transmit at the same time.
One other player advocates putting all APs on a single channel: Meru Networks. While Extricom tackles it fundamentally in hardware, Meru is primarily a software solution. “There is no standardisation of how to make APs work together – that’s the added value with a channel blanket approach. Figuring out where to place the APs and shape the antennas is the artistic part,” adds Belanger.
The advent of 11n does offer the prospect of wireless shifting from a casual, subset network in the enterprise to mainstream technology across many sectors. It is already critical in healthcare (medical telemetry, portable X-ray machines, RFID, for example).
The UK education market is also likely to be a rich seam for 11n, particularly with universities wanting to differentiate in their competition for students. At the 900-student Essa Academy in Bolton, for example, the blanket WLAN supports streaming voice, video and data applications and underpins the roll-out of Apple iTouches to each student when term begins in September 2009.
