In a couple of years, crossing the 1Gbps threshold with a WiFi access point will be routine. That access point will likely have two radios, one for each major spectrum band, and support a host of older flavors for compatibility. Eventually, WiFi will approach the robustness and speed needed to make it a completely viable replacement for Ethernet for most users.
In today’s pipeline are optional enhancements to 802.11n that have been in the works since the standard stabilized at the IEEE engineering group nearly three years ago. These enhancements will increase range and performance by up to a couple orders of magnitude, offering raw data rates of 450 Mbps and 600 Mbps.
The slated improvements will also correct for black holes, where current 802.11n gear’s signals don’t reach unless an excessive amount of overlapping devices are installed at relatively high expense. Even better, the boosts to 802.11n are just the start. A new IEEE committee is working on fast WiFi that will hit a raw encoding rate of 1 gigabit per second (Gbps).
All these higher speeds will be eminently affordable and reasonable choices for small-to-medium-sized businesses. It may even be possible to achieve higher performance (both for speed and network consistency) by spending less than a network upgrade would cost today: fewer, more powerful access points with better coverage may wind up saving money.
The need isn’t always for speed: it may be better to have a network that works in the worst circumstances, with tons of users moving lots of data, than to move additional raw data. With the popularity of watching video (for b
usiness purposes, no less), the growth in the size of standard document files, continuous network backups, and other network loads, network capacity, quality, and support for simultaneous users and heavy-load applications will become increasingly important.
Faster WiFi paradoxically also means that more wired infrastructure is needed. With individual access points able to pump out a dual-band total of several hundreds Mbps, and future dual-band devices topping 1 Gbps, more robust and higher-performance backhaul will also be needed.
The 802.11n standard has come a long way from its contentious origins, when MIMO (multiple in, multiple out) antenna arrays were seen as impractical, expensive, and beside the point of pushing data over the air. Now, in addition to 802.11n, all 3.5G and 4G cellular wireless standards either require or allow the use of MIMO for better coverage and performance.
At a time when 802.11g could only deliver 20 to 30 Mbps of real throughput out of a potential 54Mbps raw data or “symbol” rate, the idea of 150Mbps with 75 to 100 Mbps of actual throughput was pretty stunning.
But it got better. By the time manufacturers coalesced their efforts—after a cantankerous process—around a single approach for 802.11n, it was clear that all access points and adapters would support two radio streams, each of which would be able to handle a raw rate of 150 Mbps, for a combined 300 Mbps.
Each stream is a chain of radio components that share antennas. For sending, each stream transmits uniquely and simultaneously across space, using signal reflection in the environment in the same way that a billiards player uses bumpers to strike a ball. This is called spatial multiplexing: multiple signals encoded using space as a parameter. A receiver decodes the signals across multiple antennas, dumping each stream into a unique radio chain.
A receiver with a like number of radio chains, and often a similar number of antennas, can interpret the directionality of signals, sifting out separate streams to reconstruct the original message.
For instance, a 2×2 (two transmit, two receive) antenna array is often paired with two radio streams, or a 3×3 with three radio streams. Some devices with two radio streams might have 2×3 arrays, in which three receive antennas
are used to improve signal differentiation and range.
Two chains is good. But what about three? Or four? These optional enhancements to 802.11n were eminently possible, but with the exception of wireless chipmaker Marvell and startup Quantenna, most firms sat out the dance, waiting for a shoe to drop: interoperable certification from the WiFi Alliance.
The need isn’t always for speed: it may be better to have a network that works in the worst circumstances, with tons of users moving lots of data, than to move additional raw data.
The WiFi Alliance is a trade group that po
pularizes wireless networking as a technology to use while also setting certification testing for companies that want to use the WiFi label and branding on products. Apple, Cisco, Intel, Microsoft, Nokia, and many similar firms sit on the group’s board.
WiFi has managed the neat task of herding cats for a decade, even as standards have proliferated into an alphabet soup of a, b, g, e, i, aa, and more—and as the IEEE 802.11n proceedings threatened to scuttle industry-wide compatibility. The group persevered, ultimately offering a Draft N certification that was an interim brand while the 802.11n spec was being finished, and which has now transition to just “N.”
With the ratification of 802.11n as a finished spec earlier this year by the IEEE, the WiFi Alliance released an updated certification program designed to standardize several more obscure elements of 802.11n that weren’t fixed in stone until near the end.
The most important of these were multi-s
tream 802.11n devices, starting with three-stream radios. The certification process seems to have taken the brakes off the industry besides Marvell, all of which are now looking to three-stream radios and beyond. For instance, Atheros announced its 2010 series of 3×3 three-stream chips shortly after the WiFi Alliance’s certification news. The chips have to exist before the alliance can create a certification standard, which is why these evolutions all come together at once.
Likewise, while four-stream devices are defined in the 802.11n spec, the WiFi Alliance won’t be able to approve such devices until chipmakers have them available.
So far, it’s difficult to tell what hardware is shipping with 3×3 antenna arrays and three-stream systems installed. Apple’s October 2009 update to the AirPort Extreme Base Station and Time Capsule access point and network-attached storage device include three-stream radios, but Apple isn’t advertising the feature. On its base station pages, the company says only that the latest revision “gives[s] you up to 50 percent better WiFi performance and up to 25 percent better range
” than its immediate predecessor. This conforms with 3×3 antenna arrays, which, even with a two-stream radio, carry data further at higher speeds.
However, the device has the capability of higher speed, too, several industry sou
rces confirmed who declined to be identified. Apple didn’t respond to multiple requests for clarification.
The three streams in Apple’s base stations require 3×3, three-stream adapters to reach higher speeds. This means redesigned WiFi adapters in Apple gear or those made by other companies; so far, only Intel seems to have a three-stream laptop radio available, found in only a few computers.
A three-stream access point will have the ability to improve range, robustness, and throughput, but not all three at the same time. Range is nearly a given with a 3×3 antenna array, but throughput and robustness will be competing options. Let’s pick each of these apart.
Additional send and receive antennas allow an access point to push data further by providing more potential signals that a distant adapter can hear. Likewise, a distance adapter’s faint transmissions can be picked up and reconstructed better with three receive antennas. The WiFi Alliance’s technical director, Greg Ennis, said, “In a small company, the configuration of the access points geographically within the site, it’s probably more a range issue than a capacity issue.”
Broadcom, a leading maker of wireless networking chips, emphasizes range over many other factors. Mike Hurlston, the firm’s vice president and general manager for wireless LAN, said, “The advantage that a three or four stream product brings is mostly in the range.”
Hurlston said, “The interesting property of a three-stream or a four-stream product is that the third antenna or the third transmit and receive [chain] becomes a way to gather more information as a station moves farther and farther away from the access point.”
While “there’s a close-in benefit of more speed at a relatively close range,” Hurlston said that “the most interesting property to us is a much better sustainable throughput at longer ranges.”
For an upgrade over an existing network, this range improvement could mean installing access points less densely, saving money while improving performance. Or, with heavily used networks, existing access points could be upgraded and cover the same area with greater resiliency, while also upgrading bandwidth.
Adding bandwidth is, of course, useful in very busy networks with many simultaneous active users and devices, but even increased bandwidth has more impact than pure speed: any given device could move data faster across a network, which means there’s a greater amount of data that the network can carry.
“Keep in mind that the raw data rate that really gives two benefits to an installation,” said the WiFi Alliance’s Ennis. “One is that it allows for the support of applications that take advantage of that high data rate, like video. But the other is that it increases the capacity of the system as a whole.”
Ennis explained when “transmitting at a faster raw data rate, that means that they [wireless network users] are occupying less air time on the channel, which has the effect of increasing the capacity of the whole system.”
The flip side of speed is robustness: fewer dead areas in a network in which coverage is poor or unavailable, even as the rest of the network performs well, and a network that performs well even under high load or heavy bandwidth use. Using additional streams to provide redundant data through different spatial paths is what aids that process. This robustness also means that higher data rates will be available closer to an access point where, with a 2×2 two-stream system, rates can fall off to far lower than the maximum possible throughput at just tens of feet.
With chipmaker Atheros’s new three-stream chips, said Pen Li, the company’s senior product marketing manager, a couple of different technologies “will deliver more than 100 percent improvement in rate over range.” Li said that if you were seeing 180 Mbps at 30 feet, that same speed would now be available at 60 feet, instead of half that.
Li noted, “The three spatial stream [approach] itself only gives you the data rate, the throughput boost, at short range.”
The combination of consistent coverage, improved throughput, and improved range can reduce costs on the wiring side at least in one aspect: cubicles and office spaces have fewer and fewer Ethernet drops. The move to multi-stream radios might allow near-total elimination of wired connections for end users, except for those users with the heaviest data needs.
“You don’t have to wire up the cube: you can be in any conference room or any cube” and have the same benefits from the IT infrastructure, said Sameer Bidichandani, the director of technical marketing for WiFi at chipmaker Marvell. The increased bandwidth means that 10 gigabit Ethernet may be increasingly needed in network closets, however, to maintain performance across the network as end points speed up.
The move to multi-stream radios might allow near-total elimination of wired connections for end users, except for those users with the heaviest data needs.
These near-term speed increases will eventually be eclipsed by 802.11ac, an effort underway to improve data rates in the 5GHz band. Current 802.11n works in both 2.4 GHz and 5 GHz, but 2.4 GHz is overcrowded and can’t take advantage of many future improvements. The goal of 802.11ac is to push past 1Gbps, mostly using existing, well-known techniques, and improving on them.
One improvement may be in the use of wider channels. 802.11n supports 20MHz-wide channels, the same width used for 802.11a, b, and g, as well as 40MHz-wide channels, which simply double the available bandwidth. It’s far easier to use 40MHz in 5GHz, because most countries allow several or even a couple dozen 20MHz channels, which can then be bonded into a smaller number of 4 MHz channels. In the US, there are eight channels of 23 available that are widely supported and can be used to make four 40MHz channels.
Some of the current work in 802.11ac is looking at 80MHz channels, which would potentially double again the available bandwidth—or even 160MHz channels.
This would be coupled with more efficient modulation techniques. Improvements in modulation, or the process of encoding bits onto radio waves, had a more than 10 percent jump between 802.11g and 802.11n, accounting for some of the speed improvement. (The rest was through wide channels and multiple spatial streams.)
Another element being looked at is MU-MIMO (multiple user MIMO), which the WiFi Alliance’s Ennis explained would allow “simultaneous streams to be transmitted to different users on the same channels.”
None of this is etched in stone, given that the proposed ratification for 802.11ac is December 2012.
Availability of hardware based on early drafts of this technology doesn’t sound likely to hit the market until 2011, given current roadmaps, making it possible to plan to purchase three-stream 802.11n upgrades without short-term buyer’s remorse. It took roughly three years between the first MIMO devices to ship before there were plenty of draft-based 802.11n devices on the market with reasonable interoperability.
Dare I say that the wireless office of tomorrow will look much like the wireless office of today, only better? That was the promise of 802.11n, and it’s largely been fulfilled. 802.11n works better in a larger variety of office setups with less fuss to get greater speeds.
The evolution to three-and-more stream 802.11n should fill in the gaps of robustness, performance, and range that remain, giving businesses a chance to have better networks without spending enormously more.
The flexibility of three-stream devices should allow networks to be optimized for raw speed, speed-over-range, or range, without giving up much in the process.
With the next generation of standards already far along in the planning stages, it seems like we’re at an inflection point where networks will deliver enough performance in 2010 to meet most needs without breaking budgets.
Original Article By: Glenn Fleishman