Judging by the flurry of announcements in the lead-up to the recent Consumer Electronics Show, it may seem that the era of ubiquitous "fat pipe" wireless communications through ultrawideband signaling is upon us. From chip and development-kit announcements to demonstrations and partnerships, both sides in the long-running UWB debate-the Multiband-OFDM Alliance and the UWB Forum-have exposed their respective hands, making 2005 the year of truth for a technology that has long captured the imagination of the wireless, personal computer and consumer electronics industries.

Though still paying lip service to an IEEE 802.15.3a task group that has enmeshed all concerned in a Groundhog Day-style recycling of the same debate for the past two years, both camps have continued along their respective paths to market, rendering the standards work almost irrelevant. Their paths will now cross in a battle for control of the digital home. There, ultrawideband technology is aiming to dominate the wireless streaming of video and audio from set-top boxes, media gateways and PCs to high-definition TVs and laptops, as well as among cameras, PDAs and cell phones.

To do so, the two sides in the UWB arena will compete not only with each other, but also with a host of alternative technologies. Most notable among them are wireless-LAN schemes such as 802.11a/g and the high-rate 802.11n standard now under development. Other contenders are the various 802.15-based wireless personal-area networking options. At the same time, by targeting portable devices and power line networking, UWB will encroach upon ground now held by Bluetooth and even HomePlug.

At first glance, the battle between the two UWB camps is akin to David and Goliath. On one side lies the Multiband-OFDM Alliance (MBOA) Special Interest Group, led by Intel, Texas Instruments, Philips, Staccato Communications and Wisair. The MBOA is backed by an "ecosystem" of almost 200 brand-name and startup companies that form a who's who of the consumer electronics industry. It also has the support of the WiMedia Alliance.

Facing off against all that muscle is the UWB Forum. This camp is led by Freescale Semiconductor Inc., accompanied by a smattering of UWB startups and wireless veterans such as decaWave, Pulse-Link, TimeDerivative and RF Micro Devices-as well as 100 or so other members culled from industrial automation and academia.

To stake their claim in the digital home, each side must tame its own technological demons as well as those-both real and imaginary-that may be thrust upon it by the opposing camp.

Longstanding demons include regulatory compliance, both nationally and internationally, as well as complexity, scalability, cost, power consumption, interference handling and silicon availability and partitioning. These issues will come under scrutiny as the two camps seep silicon over the next few quarters.

But for many applications developers looking to get up and running with UWB, the critical decision parameters will revolve around higher-level issues. These include the quality and availability of development kits; interface flexibility, both hardware and data; and support software and application programming interfaces, libraries and drivers. In particular, given the early stages of development of UWB, chip and module vendors must be able to provide a high degree of personal attention, or "hand-holding," to potential customers.

When the Federal Communications Commission established its first Report & Order (and there may be another) in February 2002, it did so based on what was available at that time, namely "impulse UWB," upon which direct-sequence (DS) UWB-of which Freescale is the main progenitor-is founded. But since then, the MBOA has developed its own UWB flavor, based on multiband orthogonal frequency-division multiplexing (www.multibandofdm.org/presentations.html). That opened the door for XtremeSpectrum (now part of Freescale) to point out in mid-2002 that the MBOA did not adhere to the letter of the FCC's rules with regard to measuring frequency-hopped systems. Thus began the furor.

The MBOA still maintains that its technology adheres to the "spirit" of the R&O, and the FCC has thus far failed to resolve the dispute. Last year the MBOA filed for an FCC waiver to allow its technology to get out into the market so that real-world analysis could begin. But still the FCC has hesitated, pending tests now under way at ITS Labs in Colorado.

Serdar Yurdakul, director of marketing and business development at Wisair Ltd. (Tel Aviv, Israel), fully expects the waiver to pass in the near future. However, Wisair and fellow MBOA members are not waiting on the sidelines until then. They continue to press ahead with aggressive road maps to silicon based on a recently defined physical (PHY) layer. A media-access control (MAC) layer has yet to be defined.

Meanwhile, the DS-UWB camp is armed with Freescale's XS110 chip set, the only FCC-approved high-rate-communications UWB silicon. But that's only for the United States. As IEEE 802.15.3a task group chairman Bob Heile has pointed out on numerous occasions, the real issue for market success is international compliance, which requires spectrum rules that have not been forthcoming. That said, Singapore, South Korea, Ireland and Japan have set up UWB Free Zones or granted temporary waivers for testing to begin, and the U.K. is on the cusp of doing so.

Also, Task Group 1-8, the international body set up to study compatibility between ultrawideband devices and other radio communications services, has set its final U.S.-based meeting for May 18-27 in San Diego (www.uraxs.com/ITU-R_TG1-8_UWB-5th_mtg). The meeting is effectively the last chance the international community will have to settle on mutually acceptable UWB operating and test specifications that will allow that technology to operate legally outside the United States.

Although the DS-UWB camp is quick to point to the regulatory issues associated with the MBOA design, doing so may not be in its best interests. "While the waiver likely will pass, it does create more uncertainty for UWB [in general]," said Mark Moore, chief technology officer at Artimi Inc. (Cambridge, England). "The trick for us in the UWB community is to show that we can detect and adopt in close proximity and then, over distance, show that we're so low that we blend into background noise," said Jim Lansford, chief technology officer at Alereon Inc. (Austin, Texas). "It's a matter of getting people and regulatory bodies comfortable with that."

Both camps claim to be able to "notch" out in critical bands, such as 5 GHz, to avoid interfering with incumbents-and to keep from being interfered with. But Freescale contends that DS-UWB is inherently less likely to have interference issues across the board. In addition, MB-OFDM radios will need location-awareness capabilities to ensure they don't interfere with incumbents as the devices roam from country to country in places like Europe.

Silicon availability

While Freescale claims a two-year advantage over the MBOA with its 110-Mbit/second XS110 chip, the company is the only source for DS-UWB silicon, aside from emerging but noncompatible chips from startup Artimi and from Pulse-Link. And as the MBOA will quickly point out, that is problematic, given the diverse applications for UWB and the potential market size. Also, long-promised customer announcements have not materialized, even though the company did strike deals last year with module manufacturers such as Global Sun and Taiyo Yuden. It also used CES to demonstrate a UWB-enabled Motorola cell phone and myriad HDTV applications.

Equally important, the MBOA has repeatedly cast doubt over Freescale's ability to power-efficiently and cost-effectively scale beyond the 110 Mbits/s of its current implementation up to the 480 Mbits/s mandated by the IEEE 802.15.3a task group-a rate required for such applications as wireless USB.

"The primary argument for MBOA is where you put the complexity," said Alereon's Lansford. "We're putting it into the modem with a fairly simple RF. But DS-UWB with comparable performance would have to have a very complex rake receiver, which is an RF animal and won't scale with Moore's Law."

According to Lansford and his MBOA cohorts, doing the rake digitally will require a very high oversampling rate to resolve the multipath to within a fraction of a chip. "So now you have a rake algorithm for a 10-Gbit/sample system and optimized for a 32-finger rake. That's not a receiver I want to build," he said, "unless I were a university professor."

But Jerry Lynch, director of applications engineering at Freescale, rebuts the scaling argument, pointing to the company's road map, announced in June. "We're working on next-generation silicon now, and we'll see higher rates [200 Mbits/s] in the first half of 2005," he said. "The rake receiver will be digital and integrated, and that's no longer an issue. We have an architecture that can be much greater than 500 Mbits/s."

Freescale is indeed ramping up the sampling rate to approach that of the 1.3-GHz chipping rate. "The beauty of that architecture is that as you approach the chipping rate, you start to need less resolution because you can have resolution digitally," said Lynch. Lower resolution is key given that the power consumption of analog-to-digital converters goes up exponentially with resolution-but only linearly with sampling rate.

According to Lynch, the Freescale design simplifies the equalizer the rake receiver needs to fewer than five taps, vs. the 32 cited by MBOA advocates. In addition, he said, "A clever implementation keeps the gate count extremely low, because you're not replicating an entire architecture every time you want a rake finger. You're doing tricks like lookup tables, etc."

For its part, the MBOA has undoubtedly garnered widespread support, but it has yet to release production silicon, though Wisair has a demonstrable transceiver and baseband. Staccato and Alereon are not far behind in their development announcements, and all expect to release full chip sets later this year-once the MBOA's new and much-anticipated MAC layer gets defined.

"We'll have our first preproduction version at the end of Q1, though it won't be fully MBOA MAC-compliant," said Amir Freund, vice president of marketing at Wisair. The MAC will instead be a subset. In the third quarter, he said, "we will introduce a fully MBOA-MAC-compliant two-chip solution with wireless USB. It's an aggressive schedule, but we feel confident about it.".

Wisair's 501 silicon germanium-based transceiver has been demonstrated at 480 Mbits/s, but when combined with the company's 530 CMOS baseband, the highest data rate to date is 200 Mbits/s. "That's conservative, and we hope to get to around 300 Mbits/s or so-or 12 to 13 meters at 100 Mbits/s, which is a little better than we expected," Freund said.

Alereon has taken a slightly different tack with its implementation, opting for a single-chip RF and baseband. This makes sense, according to Lansford, given that the MBOA MAC is not necessary for wireless USB, "and we think wireless USB is how this thing will get off the blocks." It too uses SiGe in the front end, leveraging a relationship with Jazz Semiconductor. Alereon has demonstrated rates of up to 480 Mbits/s for its transceiver, as has Staccato Communications (San Diego).

In the end, are MBOA and DS-UWB technology really all that different?

For much of the early part of 2004, Freescale and Pulse-Link pushed the idea of a Common Signaling Protocol that would allow the two standards to at least coexist without killing each other. The MBOA would have nothing to do with it, however, and the CSM remains moribund as a concept, though Freescale may still integrate it on its chips. According to Artimi's Moore, that company is implementing a flexible architecture that can do both MB-OFDM and DS-UWB, though initial implementations will be a proprietary DS-UWB-only design that Moore said can reach peak rates of 800 Mbits/s. "Our target is to deliver 100 Mbits/s [payload] anywhere in the home using single-hop, mesh-style radios or power lines, or a combination," he said.

The company demonstrated its RTMI-100 dual-mode wireless/power line chip at CES. The chip is implemented in 0.18-micron CMOS with a SiGe front end and packs most of the low-noise amplifier, all the analog, baseband and MAC, and all the back-end I/O on a monolithic slab of silicon. Three separate RISC cores cater to the PHY (with digital rake receiver), MAC and I/O processing, respectively. The company expects to announce silicon availability this month for its DS-UWB design and expects to have a dual MB-OFDM/DS-UWB version by year's end.

Getting going with UWB

For many developers ramping UWB, the first thing, according to Alereon's Lansford, is to be careful about getting a flat response from the antenna, given its broadband requirements. Aside from careful interfacing to and matching of the antenna, Freescale's Lynch points to I/O matching in consumer devices and to software. It's here where he believes Freescale has an advantage with an emphasis on TCP/IP.

"TCP/IP has a huge applications base with universal OS support, vs. USB, and we want our customers to focus on the application, not the interface," he said. Wisair's Freund also pointed to the WiMedia convergence layer, which allows MBOA silicon to cater to Universal Plug and Play, wireless USB and 1394 interfaces. All UWB vendors are emphasizing APIs, libraries and the requisite hand-holding customers need for new technologies.

However, developers that don't want to get their hands too dirty with UWB can follow the path of Focus Enhancements, which contracted with Cadence Design Systems Inc. (San Jose, Calif.) to get its video projector systems UWB-enabled.

"One of the biggest issues today is a very low tolerance for risk, and we've been in design services for many years and so have developed a very good track record to reduce our customers' risk," said Tim Henricks, vice president of engineering services at Cadence. "We've developed a methodology and flow to give a very high probability of first-pass success. For a startup whose success depends on this coming right out and working, this is very important."

Otherwise, designers can get rolling with development kits from the various vendors that are trickling onto the market. Freescale's UWB-EVK is available for $399. Wisair has also rolled out a kit, the DV9100; the price was not available by press time. Alereon's new AL-4000 EVK costs $15,000 each (a pair is required) and will see limited availability this month.