Manhasset, N.Y. — On a conceptual level, the announcement last week that the Bluetooth Special Interest Group would work with ultrawideband proponents to make its technology compatible with UWB is a win-win collaboration. Bluetooth gets a high-speed interface, UWB an evolved ecosystem. And consumers get the benefits of both, with faster time-to-market.
But many questions arise on an implementation level, not the least of which is what aspects of Bluetooth will migrate to UWB and how will they interoperate with the media-access control (MAC) layers already defined by the WiMedia Alliance and the UWB Forum, the two highly competitive UWB groups on distinct parallel paths to market. Also unknown is what becomes of the "old" Bluetooth physical layer (PHY) — and its manufacturers. Finally, some are wondering whether the Bluetooth SIG might have bet on a nonstarter, given that UWB at the moment is only allowed to operate in the United States.
Beneath the surface lie more-subtle questions surrounding the future of Wireless USB, the interface protocol highly promoted by the WiMedia Alliance, which may now cease to exist. Also in doubt is the role of CWave, a third UWB PHY now under the wing of the UWB Forum.
Bluetooth at crossroads
Whatever the questions, there was no doubt in the mind of Mike Foley, executive director of the Bluetooth SIG, that something had to be done to ensure Bluetooth's longevity with a high-speed road map. "We debated if we should develop a higher-speed PHY but decided the collaborative approach was better," he said.
Even with its Enhanced Data Rate (EDR) specification, Bluetooth reaches only 3 Mbits/second. That is sufficient for voice and low-data-rate applications, but video and the gigabyte downloads expected in the future would require much more. "Members asked what would come after EDR," said Eric Janson, vice president of CSR North America, whose parent company, CSR plc (Cambridge, England), developed the first single-chip Bluetooth offering. The idea of a 10-Mbit/s interface was tossed around early on, "but we saw competition with various other protocols [such as wireless LANs], and we didn't want that," Janson said.
With the UWB collaboration, Bluetooth now potentially jumps to 480 Mbits/s, leapfrogging those other protocols. "Bluetooth will operate as a signaling channel and provide all the profiles and interactivity and open up UWB as a fat pipe," said Janson. According to Foley, UWB will also benefit from the "heavy lifting" the SIG has done to educate consumers on how to use short-range wireless interfaces.
The enthusiasm of Janson and Foley is matched on the UWB side. "It's excellent," said Martin Rofheart, director of UWB operations at Freescale Semiconductor Inc. (Vienna, Va.). Describing UWB as a "mirror image" of Bluetooth, he said that little has been done to date to ensure ease of use or infrastructure for testing and certification. But high data rates are readily available, he said, and both technologies promote low-power operation.
For WiMedia, it's "mission accomplished," said Stephen Wood, president of the WiMedia Alliance and technology strategist at Intel Corp.'s Communication Technology Lab (Hillsboro, Ore.). "It's part of a long-term strategy we've been cultivating. We announced over a year ago the common platform, and this is probably the last critical element we're likely to add to it." The platform now supports Wireless 1394, Wireless USB, Internet Protocol and Bluetooth.
For one analyst, the marriage was something of a vindication. "I told you so!" said Craig Mathias, principal at Farpoint Group (Ashland, Mass.). Mathias has long predicted the demise of Bluetooth as a distinct radio operating in the 2.45-GHz band. "Now that it's separated from a physical layer, it'll be fine," he said last week. As for UWB, "it is likely to become the short-range, WPAN [wireless personal-area network] radio of choice for a broad range of applications."
But Dan Benjamin, senior analyst at ABI Research (Oyster Bay, N.Y.), was more cautious. Benjamin expressed concerns over Bluetooth's need to support backward compatibility as well as the two current UWB PHYs. "That's going to increase cost by two to 10 times," he said. "It's too early to say what this means." Benjamin believes the SIG will have to decide between WiMedia's multiband-OFDM PHY and the UWB Forum's direct-sequence PHY.
While the idea of using Bluetooth for command and control is sound, just how that is to be done remains vague. To date, the SIG has outlined a "convergence layer" that would be used to interface between one of the two UWB MACs and Bluetooth's logical-link control and application protocol layer. The Bluetooth radio will handle audio, using its well-proven synchronous connection-oriented protocol.
"It [Bluetooth] will be a protocol that can be added to many devices and provide features such as device and service discovery and profiles," said Foley of the SIG. While Bluetooth's profiles have long been its Achilles' heel and have slowed its progress, CSR's Janson said that up to 20 have now been nailed down and "they're getting better and better."
Device discovery is an especially important feature for the UWB camp, said Janson, particularly in light of what he called a "deficient" Wireless USB implementation by WiMedia. Device discovery allows two wireless devices to wake each other up, perform recognition and establish a secure connection. According to Janson, the Wireless USB design requires that the user first make a wired connection using a USB cable before going wireless, which defeats the purpose of being wireless in the first place. "It's not very well thought out — it will be dropped with this [collaborative] effort," he said.
According to Mark Bowles, vice president of business development and marketing at Staccato Communications Inc. (San Diego), the wire-based association model is being considered, but only as backup. "It might be mandatory in the standard, but it will not be the primary association model," he said. "It's a very complex issue and depends a lot on the input capabilities of the device, whether it has a display, keyboard, etc., or not." The group is in the process of defining the model of choice and while there may be some overlap with how Bluetooth does the job, Bowles likes the 802.11 model. "It would be nice if we all had one common method," he said.
The migration to UWB puts CSR in an interesting position, given that its bread-and-butter has been single-chip Bluetooth radios in the 2.45-GHz band. However, the need for backward compatibility means that traditional Bluetooth radios won't disappear any time soon, said Yoram Solomon, director of strategic business development in Texas Instruments Inc.'s Connectivity Solutions Group, which makes both Bluetooth and UWB radios. Moreover, TI's interest is mainly in handsets and Solomon believes Bluetooth will still be used for headset-to-handset connection.
However, Solomon sees UWB as also being mandatory in handsets due to the increasing amounts of storage going into them to handle image and video data — and the need to share that information quickly. According to Steve Nelson, vice president of the connectivity group at Standard Microsystems Corp. (Hauppauge, N.Y.), the lowly IrDA interface is the best way to achieve this right now (see related story, page 46). TI's Solomon disagreed, pointing to the introduction of 512-Mbyte MultiMedia Cards for handsets. "Can you imagine trying to download that at [IrDA's] 4 Mbits/s?" he asked.
The co-location of Bluetooth and UWB on handsets raises interesting integration challenges. Both Janson and Solomon see a scenario whereby they will be implemented separately at first, then migrate toward a single-chip, software-radio-type design with a single antenna. That antenna design will have to accommodate the bands from 2.45 GHz, for Bluetooth, right out to 3.1 to 10.6 GHz for UWB. "That has to be worked on," said Solomon.
In the meantime, CSR is developing its own UWB radio, although Janson said it has not yet decided which PHY to focus on. "But when a standard is ratified by the SIG, we'll have chips ready," he said.
While the SIG is working with both WiMedia and the DS-UWB Forum, "ultimately there will have to be a single high-speed radio," said Foley. WiMedia's Wood agreed. "The matter will resolve itself going forward," he said. "It's up to the Bluetooth SIG to decide."
The addition of a third PHY, CWave, to the UWB fray may not confuse the issue all that much, said Bruce Watkins, president of Pulse-Link Inc. (San Diego), which developed it. Watkins said the PHY works under the same MAC as defined by the UWB Forum, so it should interoperate with whatever the convergence layer defines.
Regulations a hurdle
The criteria for deciding on a UWB PHY depend to a large degree on the packet size and power consumption of each, as well as its ability to meet international regulations, once they're defined, said Foley. Right now, the United States is the only nation to have rules for UWB in place. The International Telecommunications Union, Radio (ITU-R) will meet this week in San Diego to gather information on UWB (uraxs.net), but the group is a long way from agreeing on a spectral mask that defines the power limits and operating bands allowed for each region of the world. As a result, it's possible that UWB's — and now Bluetooth's — potential may never be realized.
On the other hand, the migration of Bluetooth to UWB may "lower the barrier for acceptance of UWB" by the international community, said Foley. That is also an argument put forth by proponents of IEEE 802.15.4a, a new standard in development that will take advantage of UWB's ability to do both ranging and communication (see story, page 55).
