The announcement of a Wi-Fi alternative for fast Bluetooth shouldn't have been a big surprise, but it breathed a chill over the Bluetooth Evolution conference in London today, and provoked a reaction from ultra-wideband vendors.
In fact, when the Bluetooth Special Interest Group (SIG) announced plans for a fast Bluetooth on UWB in 2006, there was always room for more technologies. UWB is an alternative transport, that's meant to be implemented alongside regular Bluetooth and under its control. Only when fast transfers are needed will the regular Bluetooth switch on the ultra-wideband.
The Bluetooth SIG has already endorsed another alternative transport - to be based on Nokia's Wibree - but the Wibree-based ultra-low-power (ULP) Bluetooth is so different from UWB that there was never any danger of confusion.
Putting Bluetooth on top of Wi-Fi is likely to be more controversial - or at least difficult for the Bluetooth SIG to explain - as it offers an alternative fast Bluetooth migration. "If the SIG is to serve the market, we have to be careful how we roll out our positioning," said John Barr, chair of the SIG's board of directors, and director of standards realisation at Motorola, when he put the Wi-Fi option front and centre at London's Bluetooth Evolution conference.
"If UWB can produce something acceptable, we will use it," he said. "As SIG chairman I can see where UWB can be used right away - other than on phones," he added. On phones, however, the technology is not yet delivering, he said - while there are many dual-mode Wi-Fi phones already out there and on their way to the market.
Overlap, not competition
In fact, UWB and Wi-Fi are overlapping technologies, rather than competing ones. Wi-Fi has a longer range, and a public infrastructure, while UWB has a shorter range, and a great speed-and-power profile.
"UWB will be able to transfer a 900 Mbyte two hour movie in 22 seconds," said Glyn Roberts, president of UWB standards group, the Wimedia Alliance. High-definition movies, at 18 Gbyte, would transfer in seven minutes, he said. In one second, UWB will be able to transfer a whole album in MP3 format or twenty high resolution photos.
By contrast, Wi-Fi currently has a top actual transfer speed of around 22 Mbit/s using 802.11g, which is likely to be the fastest Wi-Fi on handsets for some time, according to some. "It's a medium speed technology," said Nick Hunn, chief technology officer of Ezurio, pointing out that the higher speed options within 802.11n rely on multiple antennas which are very hard to build into a handset.
But medium speed might well be enough for now, says Barr, who points out that it may be all a phone can handle in any case. "The maximum data rate on a phone will be less than 180 Mbit/s for some time," he said. "These are not PCs. We don't build them with a wide data path that chews up power." Right now, he could care less about the "mythical" 480 Mbit/s.
So on this view, till phones get a substantial re-engineering, UWB-enabled handhelds are a fantasy.
Is Wi-Fi really a quick fix?
Till that happens, handset makers say that Bluetooth on Wi-Fi would make sense, given the number of Wi-Fi radios there are in new and upcoming handsets. "We want to use the radio that is already there," said Patric Lind, senior specialist in local connectivity at Sony Ericsson.
This sounds attractive, especially given the very low attachment rate of Wi-Fi in handsets. While everyone can use the cellular radio, and a lot of people manage to get the Bluetooth working, very few make regular use of Wi-Fi because it is tricky: a Bluetooth top layer that made handheld Wi-Fi easier is a tempting idea.
But how quickly can Bluetooth and Wi-Fi be melded? The technology spec won't be complete till next year, possibly as late as the third quarter, according to John Barr. "I'd expect phones six months after that."
Since the new implementation would be in software, it's technically possible that it could be implemented on existing dual-mode phones, according to Cambridge Silicon Radio's Robin Heydon: "It's a generic MAC and PHY, so we don't need to change the Bluetooth silicon. It could be a software upgrade."
"That's not likely," said Hunn. "The concept of Bluetooth on Wi-Fi has got legs, but we're not sure how many and which way they're walking." Apart from the technology issues at baseband level, Wi-Fi has evolved a different security model, designed around clients and access points, while Bluetooth is a peer-to-peer protocol.
Any shift to Wi-Fi is a short-sighted, conservative, European (and presumably US) viewpoint, ignoring the reality of UWB, according to Mark Moore, chief technology officer of UWB vendor Artimi: "There will be UWB handsets in Asia, within the next six months." One such was announced by SK Telecom of Korea earlier this year. Handset makers shouldn't grumble about delays in the technology, he said, since the chips will be available well before the completion of the Bluetooth-on-UWB standard.
UWB products in the market use the Wireless USB variant of UWB - which has around 29 products certified. Bluetooth SIG people quietly dismiss W-UWB as being put together by PC engineers, not "real" radio people. Their UWB-based Bluetooth will be a completed standard, with three prototypes by December this year, they hope, with silicon to follow.
Timing is all?
UWB vendors may dislike this change of focus by some in the SIG, but it may just be a question of timing. Even Barr believes UWB will ultimately win out: "I'm not de-emphasising UWB," he said. "We know we need higher speeds, and our architecture can integrate it - it's designed to be independent of the radio."
In the end, it's ironic that Wi-Fi is seen as a conservative option, by some in the SIG, says Pierre-Yves Couteau, director of strategic marketing at Philips spin-off NXP: "In 2000, if you'd said you were putting Wi-Fi on a phone people would have said you were crazy," Today's UWB skeptics will see that technology also become mainstream, he said.