One size fits all

GUEST OPINION: IEEE 802.11 WiFi is a technology that has taken a decade to mature and fit into a particular market segment.

By Justin Howard, Staccato Communications -- EDN, 5/6/2009

You'd have thought with all the press lately that WiFi was the solution to everything—universal technology capable of being put in everything from toasters to automobiles. The reality is that 802.11 (as it's officially known) is a technology that's taken a decade to mature and fits quite squarely into a particular market segment.

Of late there's been a lot of bupkes created by a single carrot; 802.11 silicon is getting so cheap it must be able to interconnect the world with breakneck throughput.

Historically speaking

When I started developing WiFi products over a decade ago, things were very different. Barely anyone had heard of 802.11 at the time. And its friend the WiFi Alliance (a sort of ad hoc collection of vendors looking for compatibility between equipment) was the merest glimmer of a thought bandied around the bars after an 802 standards conference.

Back then WiFi was fighting it out with Intel's HiperLAN and HiperLAN/2, and no one really had an idea who'd win. Things were a mess. 802.11a wasn't getting any attention due to the 5-GHz radios and spectrum licensing issues. 802.11b was working well in the unlicensed 2.4-GHz band but could only muster 2 Mb/s. Bluetooth was just around the corner and set to cause mayhem for the next decade.

No one could imagine a wireless world that wasn't dual-bonded GSM lines at 9.6 kb/s let alone capable of the heady hype we hear today. WAP was our best wireless (network) technology and only moderately more fun than bathing your grandma.

If we move forward a couple of years, 802.11b matured up to 11 Mb/s and people were vying for the next big speed jump, 54 Mb/s. Half the people returned to 802.11a, which was built for this purpose in 5 GHz, and the others attempted to crowbar OFDM into 2.4 GHz by adapting 802.11b's DSSS to 802.11g. (For the humbled reader, OFDM and DSSS are just ways of sending data wirelessly. One provides better throughput over the other but at a higher cost in silicon.)

When the hubbub died down, we were left with a confused market of mixed-mode 802.11a/b/g devices and a wealth of incompatibility that demanded that the fledgling WiFi Alliance step up its game.

Learning from our mistakes

Now you may be wondering why you've just read a brief history of time, but the echoes of history resonate well in ultra wideband (UWB). It, too, is a technology that's fought for maturity and is now fighting for its place in history. The only difference is that it's been planned better by learning from the mistakes of the past.

Within our context, UWB actually means a collection of things beyond simply being an ultra-wide radio band. Firstly, it defines a physical radio standard known as WiMedia. This is quite important to grasp as, unlike 802.11, care has been taken to separate on-air data transmission from any overlying technology such as USB.

The concept allows for flexibility in how a technology is constructed and means that whether Wireless USB or Bluetooth there's no conflicting coexistence problems or pollution caused by biasing a particular camp.

Additionally, WiMedia has always been a timeslot based architecture where devices make reservations for air time; this gives a number of superior qualities to 802.11:

• Zero collisions
• Low power
• High throughput

In a way, these are the base elements of UWB and the key to its future success. Many technologies can vie for product space, but these three elements combined with a devolved radio technology are unsurpassed.

Zero collisions

This isn't strictly true; it's more of a half-truth. WiMedia will very quickly adapt to such interference and accommodate radio newcomers. Conversely, older 802.11 devices make no attempt to balance their network in this way and can easily break transmissions on newer devices, particularly at range. The narrow 2.4-GHz band is also susceptible to higher levels of interference.

When something hasn't been acknowledged it's retransmitted. When it's received badly it's retransmitted. When two or more computers communicate at the same time and can't hear each other (but can hear a router) it's retransmitted. Pretty much anything can screw up transmission, and if it happens too often, the transmission rate between you and your router will also automatically be lowered to try and compensate for the problem.

Most of the time it doesn't, but the costs is high. 2.4 GHz is a crowded, narrow frequency band used for more than just 802.11 radios. This leads to a category of interference problems known as coexistence.

Bluetooth is probably the most known 2.4-GHz neighbor; however, due to its being the preferred mechanism for audio headsets, it's also the most destructive. A phone making an 802.11-based Internet call using a headset creates a paradox: Both technologies are needed at the same time, but neither technology knows the other one exists.

The result? Collision after collision and your 802.11 data rate lowered to the minimum (anything from 2 to 11 Mb/s, depending on whether 802.11 a/b/g/n).

If Bluetooth and WiFi were based on UWB and shared WiMedia as the radio protocol, coexistence wouldn't be a problem. Indeed, WUSB and UWB Bluetooth do exactly that. Sadly though, even now in 2009, people don't make Bluetooth headset-based Internet calls over 802.11, as it just doesn't work.

Low power

So what of power, and why is this important? Power is paramount to any type of portable device; it's a precious resource that is rationed to ensure it will function as long as possible.

WiMedia's timeslots mean a device can sleep very accurately, only wake-up when needed, and, hence, save power. For 802.11, power control is much harder. Timing clocks are much less accurate and more time is spent awake ensuring everything is synchronized.

802.11 does have power-saving and transmit arbitration features, though, but these weren't introduced until much later. It means support for it is patchy within routers, and this adversely affects the battery of a portable device when using routers without its support.

In the end, 802.11's problems with collisions and coexistence also compound the problem. Yet more time is spent awake resolving conflicts; combined with its extended range and slower transmission speeds, this results in it using substantially more power than UWB across its broad spectrum.

High throughput

Finally there's throughput. When comparing headline figures, 802.11n (the higher-data-rate, longer-range extension to 802.11) looks comparable to if not better than UWB. But unsurprisingly, the design goals of 802.11 and UWB mean this isn't an accurate comparison.

UWB is a short-range technology capable of high data rates and low power. 802.11 is a long-range technology capable of high data rates if the conditions allow and power permits.

UWB uses its high data rates and short range to actually save power. 802.11's battle with co-existence, timings, rate selection, range, and, finally, backward compatibility works against any power-saving or high-speed data rates.

In the end, when UWB is put into a Wireless USB product, that data rate exceeds 150 KB/s of real data throughput. Data is transmitted over the air at 480 Mb/s, with the losses incurred by the USB protocol itself. Yet with aspirations for transmission rates to increase to 2 Gb/s and aggressive power saving when not sending, UWB outclasses 802.11 on its transmission power per bit by a large margin.

Conversely, 802.11, at least on paper, is creeping into UWB's domain. However, the biggest Achilles' heel of the standard is its backward compatibility. Beacons[1] have to be transmitted at 1 Mb/s for 802.11b network support, and this steals a large amount of the available capacity from the network. If running a mixed-mode system, it's impossible to achieve the headline figures promised by 802.11n. (A beacon is how a router advertises it exists. This typically happens 10 times per second and contains critical information about the 802.11 capabilities supported by the router.)

Travelling the road

Historically you can see how UWB has learned from 802.11 and found a clearly defined niche for itself solving real-world problems. It's important to note these aren't competing technologies, they're complementary technologies.

802.11 is a mature, mainstream technology perfect for networks and the Internet, giving accessible range and speed. UWB, however, is a peripheral technology that targets portable devices and delivers excellent battery life, throughput, and integration.

To consider expanding 802.11 beyond its bounds would ultimately lead to unsurpassable limitations and an inferior end-user experience. The hyped rumors have been generated due to the price of silicon and a polarizing of UWB as weaker players have left the game. 802.11 has historically fought battles; the problem, however, here is that there is no common ground with UWB to fight for.

One size doesn't fit all.