An IEEE 802.15.3a task group in San Francisco will evaluate the Multiband-OFDM Coalition's proposal and numerous others. A thicket of issues ranging from technical merit to royalties, time-to-market, FCC compliance and market positioning will likely churn with the inevitable member politicking, rendering the outcome anybody's guess.

But growing pressure from both consumer electronics and PC companies could bring a quicker decision on ultrawideband (UWB) than other IEEE standards efforts have received, especially those for 802.11, on which Wi-Fi wireless LANs are based.

"The TI and Intel merger is a powerful force that will form the critical mass to really set UWB on track," said Martin Reynolds, an analyst who examines future technologies for Gartner Dataquest Inc.

The IEEE 802.15.3a group will have to decide on a standard that allows for high-speed (up to 480 Mbits/second), low-power, short-range wireless communications among multiple connected devices. Dropouts and merged proposals have compressed an original field of more than 30 proposals from a who's who of semiconductor PC and consumer electronics companies since the February submission deadline.

"This is going inordinately fast through the IEEE process vs. 802.11," said Stephen Wood, strategic-marketing manager with Intel's R&D division. "What's different about it is that we were able to get a group of people to work cooperatively very early-both on- and offline-so this has accelerated the process."

The Multiband-OFDM Coalition (see www.eet.com/story/OEG20030714S0060) brings TI under the same umbrella as the 14 members of the original Intel-led UWB Multiband Coalition, thinning the field of ultrawideband proposals. This leaves the partnership of Motorola Inc. and longtime ultrawideband proponent XtremeSpectrum Inc. (XSI) as the staunchest opposition to the new Intel/TI alliance. Independent proposals, including those from Mitsubishi, Oki Electric and STMicroelectronics, are also being considered by the .15.3a group.

Despite Motorola's established history in wireless communications, Reynolds said, "In the face of monsters like Intel and TI, it's not the giant it once was." As a result, the Motorola/XSI partnership will rely on XSI's advantage of having ultrawideband chip sets available now, giving it a leg up in the standards process.

"TI and Intel's proposal is paper, simulations, analysis," said Chris Fisher, vice president of marketing at XSI. "Won't see silicon for two to three years. Intel is still trying to build 802.11 silicon, for God's sake, never mind something as bleeding edge as UWB." Fisher said consumer companies are pressing for products today. Five companies-Sony, Philips, Samsung, Panasonic and Sharp-will express that interest in a presentation this week, he said.

That's not enough, Reynolds said. "There are billions at stake here. For a technology as important as this, they have to make it zero [royalties]."

A single PLL will generate multiple center frequencies in the Multiband Coalition's proposal.

But a TI spokeswoman did not rule out royalties, saying only that the Multiband Coalition would meet the IEEE's requirements for fair and reasonable access to its proposal's technology, then evaluate its options with the goal of ensuring the technology's rapid adoption.

Chip sets that support the IEEE's chosen standard would reside in future communications equipment, mobile gear, consumer electronics systems and computing systems of every stripe. With so much at stake, a struggle over the final standard was inevitable. XSI's Fisher said it boils down to a battle between consumer electronics (CE) companies and PC companies.

"This is a technology that has huge potential for the CE market," he said. "CE companies extract their profit and value by building most of the silicon themselves. Their concern is that the [ultrawideband] connectivity can be used as a bridge to connect the PC platform-with the Intel processor and the Windows OS-and allow it to segue into consumer devices." But CE companies are pushing back "because the likes of Sony, for example, doesn't want to be diminished to the level of Dell. Same for Samsung. That's not their business model."

Intel disagreed with Fisher's view. "UWB has a strong CE flavor, true, but we see the cellular, PC and CE markets merging. So, we'll accommodate the needs of each of those," said Intel's Wood. "It's mainly a MAC [media-access control] issue and we're not at that stage in the standardization yet, at least not at the same intensity the physical layer is being tackled right now."

Analyst Reynolds said Intel doesn't care about wireless, per se, and is strictly in the business of enabling high-rate wireless connectivity to get its computing platforms connected at megabit rates. "Note their recent fixed-wireless announcement," he said, referring to Intel's plan to develop silicon for the IEEE 802.16a standard for broadband wireless access and Wi-Fi backhaul. "The computer side was the most profitable for them this quarter, and they want to drive those profits."

Intel and TI will get to show how serious they are about ultrawideband this week as the .15.3a officially begins vetting proposals. Among the criteria under which the coalition's proposed architecture will be evaluated are power consumption, range, data rates, complexity, interference avoidance, multipath avoidance, number of connections supported, size, cost and potential time-to-market. Anuj Batra, a member of the technical staff at TI's DSP Solutions R&D sector, said the coalition's proposal is up to the task. Batra will present the proposal to the .15.3a group this week.

To address time-to-market, complexity and regulatory issues, the group chose a multiband approach that divides the available spectrum (3.1 to 10.6 GHz) into seven 528-MHz-wide bands, with center frequencies of 3.432, 3.960, 4.488, 6.336, 6.864, 7.392 and 7.920 GHz. The lower three bands are mandatory, while the upper four-above the 5-GHz U-NII band-are optional, said Batra. "This allows interoperability with the original seven-band proposal from Intel and the Multiband Coalition," he said. "Although we have also defined bands within the 5-GHz band, we reserve them for future use."

The choice of center frequencies was not accidental, said Geiganesh Balakrishnan, also a member of the technical staff at TI's DSP Solutions R&D sector. Aside from allowing rapid switching, "we can generate all the frequencies from a single PLL [vs. many individual phase-locked loops], thereby saving on power," he said.

From a complexity, multipath and basic energy-gathering point of view, Intel liked TI's use of orthogonal frequency-division multiplexing (OFDM), said Wood. "TI had a significant advantage with respect to energy capture, giving us a more robust link and/or getting more range," he said.

"With OFDM we can collect 98 to 99 percent of the energy, giving us a 3- to 6-dB improvement [over non-OFDM designs]-depending on the path-loss models," Batra said. The proposed interface can reach rates of 480 Mbits/s at distances of up to 2.5 meters, he said. (The .15.3a desired specification is 480 Mbits/s at 1 meter.) "OFDM also has a nice upgrade path to higher data rates, which is nice," he added, referring to rates of 1 Gbit/s and above. TI chip sets based on the Multiband Coalition's ultrawideband implementation will appear in 2005, he said.

Also, Batra said, because the power doubles for every bit added to the converter's resolution, fewer bits translate to lower power consumption. Batra expects the complete receiver and transmitter digital portions to occupy no more than 295 kgates. While both Intel and TI said they have worked closely with the Federal Communications Commission to ensure their multiband proposal meets FCC requirements for spectrum use and power, XSI's Fisher said he doubted their ability to do so and still reach .15.3a's desired data rates. Fisher contends that the coalition's multiband approach is essentially a form of frequency hopping. XSI's proposal takes a single-band approach.

Coalition members said they are aware of the requirements. "We obviously recognize all the regulatory questions of the FCC," said Wood. "We've checked with them to get their opinion before we proceeded and they said 'yes, it looks good.' We wouldn't have made this jump unless we weren't relatively confident it could work."

TI was more precise. Batra, Balakrishnan and others at the company said there's no requirement in the FCC's Report & Order document that the frequency hopping must be stopped when making measurements for average power compliance. Further, a three-band Multiband OFDM system has a one-over-three duty cycle on each individual band to achieve the allowed average transmit power of -41.25 dBm/MHz. Therefore, they said, there is no inherent data rate limitation for the multiband OFDM system.

But Julius Knapp, deputy chief of the FCC's Office of Engineering and Technology, said commission rules "clearly state that frequency hopping must be stopped when taking measurements; it doesn't matter if it's peak or average power measurements," he said, citing FCC documents.

XSI's Fisher said the Multiband-OFDM Coalition "[will] have to petition the FCC for a rule change, and that's going to delay the [.15.3a] effort even more."