BBH Central IconBBH Report Home PageSandy and Dave
  CENTRAL home  |   REPORT home About/Contact Us  |   Subscribe  |   Index by Topic  
The May 14, 2006 Issue Provided by System Dynamics Inc.
Table of Contents Print this article Email this article to a friend

Ultra Wideband -- Real Applications Coming Soon

More than four years after the FCC issued an order setting the rules for ultra wideband (UWB), volume products are close to market. We think this will have a major impact on the broadband home, and are taking this opportunity to update our readers on what is coming.

In Ultra Wideband (UWB): An end to all those cables! (BBHR 1/24/2005), we wrote "UWB advocates agree on the goals: to create "personal area networks" connecting many devices together at very high speed. These goals are similar to Bluetooth...but operating at a much higher speed. Picture a PC and its peripherals (keyboard, mouse, printer, scanner) plus digital cameras and camcorders all connecting together automatically whenever they come in range of each other."


UWB uses a very weak signal over a very wide frequency band; more familiar wireless technologies send a strong signal over a much narrower band (Wi-Fi uses 20 MHz channels). The FCC's 2002 order established a "spectral mask" that defined the maximum emission limit across the spectrum. The order opens a very wide band (3.1 to 10.6 GHz) for use by UWB; requires that transmitters occupy at least 500 MHz at any time; and sets an emission limit so low that UWB looks like noise to other users of the same spectrum. Because UWB works at such low power, it only operates close-in, typically within a single room.

UWB is not a single technology--there are many ways to modulate signals and stay within the FCC's rules. Many approaches have been proposed, and at least three distinct technologies have been brought to the chip level with demonstrations of devices and applications. All share the goals of providing very high speeds (100 to 480 Mbps) over very short distances.

In early 2003 the IEEE formed Task Group 802.15.3a to resolve the conflicting proposals and come up with a single standard for "personal area networking" (PAN) based on UWB. The task group moved quickly to reduce many proposed technology alternatives to two proposals. As we discussed in the earlier article, the companies backing these approaches grouped into several competing organizations. The Multiband OFDM Alliance (MBOA) and the WiMedia Alliance favored an approach called MB-OFDM, while the UWB Forum favored DS-UWB and had working chips and prototype products at CES in 2005. The IEEE process drifted into a deadlock, with neither camp able to muster enough votes to resolve the conflict, but each having enough to block the other. Early in 2006, the IEEE task group gave up, and voted itself out of existence.

With the breakdown in the standards process, the technology players pursued their own courses. Freescale Semiconductor continues to promote chips based on DS-UWB. Pulse~LINK, one of the UWB pioneers, recently announced a UWB chipset operating at Gigabit per second speeds, with a single chip supporting communications over the air, on coaxial cable and over power lines.

But most of the activity has been with MB-OFDM. In 2005, the two MB-OFDM groups joined forces under the WiMedia Alliance name. Seeing that the IEEE process was unlikely to result in a standard, the members started working to bring to market products based on their draft proposals. These first WiMedia products are expected to reach the market later this year.

WiMedia has attracted many key players in the IT and CE worlds. The "promoter members" include household names like Hewlett-Packard, Intel, Kodak, Microsoft, Nokia, Philips, Samsung and Sony.

WiMedia has worked hard to draw support from other industry associations. Organizations representing USB, Bluetooth and 1394 technologies have all endorsed WiMedia's UWB as the high-speed close-in radio technology.

Certified Wireless USB

Certified Wireless USB will be the first volume product incorporating WiMedia UWB. Its stated objective is "to preserve the functionality of wired USB while also unwiring the cable connection and providing enhanced support for streaming media CE devices and peripherals." Its "performance is targeted at 480Mbps at 3 meters and 110Mbps at 10 meters." Other UWB technologies could be combined with USB, but only WiMedia UWB has been endorsed by the USB Implementors Forum (USB-IF), which promotes USB and tests and certifies USB products for compliance and interoperability.

Any device that now uses USB 2.0 could be upgraded to add wireless USB as well. This is especially attractive for devices that are normally used untethered - such as digital cameras, MP3 players, and remote controls. We have often found ourselves on the road and discovering that we've left the special USB cable for our digital camera at home; wireless USB would remove the problem since both the camera and the laptop would have wireless USB built in.

There are more than 2 billion legacy wired USB connections in the world today. All new systems are equipped with USB 2.0 which operates at 480 Mbps. Wireless USB runs at the same speed and uses all existing USB firmware so a device already equipped with USB 2.0 can be easily upgraded: the manufacturer can leave the wired connection in place and add a WiMedia UWB radio.

For an update on wireless USB, we talked on the phone with Jeff Ravencraft, Chairman/President of USB-IF. At CES in January 2005, Jeff gave a talk about wireless USB "Wireless USB Initiative: First Hi-Speed WPAN Interconnect" (Adobe Acrobat PDF, 4.7 MB) in a panel Dave moderated. At that show, we saw prototype UWB products and heard predictions that products would reach the market by the end of 2005. In January 2006, we again saw many prototype WiMedia and wireless USB products at CES, and heard similar predictions for 2006.

We asked Jeff why it is taking so long to get products to the market. He said USB-IF was following a very deliberate path to market: "We're following the exact same model as USB 2.0. We know this story better than anybody. We're not going to market with pre-standard products--we're certifying right from the get-go so when the consumer buys one they know it will work. The products will be compliant to the specs - and interoperable - they're all going through the compliance/certification program."

Others have told us that initial certified products will reach the market later this year, perhaps as early as this summer, and that CES 2007 would have many real products. Jeff agreed that "CES will be a big opportunity." He thought the market would respond very positively: "This is a no-brainer for people--the user models are already there, consumers are already educated about being wireless and untethered. The irons are hot and the marketplace is ready."

At the recent CES, we saw several demos of wireless USB "dongles": devices that plug into a USB port and provide a wireless USB connection. Jeff said initial products will include host wire adapters (HWA) that plug into external USB 2.0 connectors on PCs, and adaptors based on PCI and PCI Express that are installed inside PCs. There will be device wire adapters (DWA) that connect to USB ports on devices such as digital cameras, and will also be built into USB hubs.

We asked Jeff how quickly wireless USB will be built into new PCs. He didn't want to be specific: "in my day job I work at Intel, so I won't make product announcements." But he suggested that PC manufacturers would start to incorporate PCI Express wireless USB cards in new PCs "before year end" and start to include UWB radios on motherboards during 2007.

UWB is currently approved by regulators only in the US. We asked Jeff how quickly UWB will move into markets outside the US. He pointed out that although the IEEE standards process reached a dead end, ECMA International--an IT-centered standards association based in Geneva--has issued two standards for UWB based on WiMedia. Jeff said ECMA submitted these standards--ECMA 368 and ECMA 369--to the ISO in January, and they are being "fast tracked" as global standards with approval expected this year. Both Japan and the European Commission are working on spectral masks for UWB, with approval expected before the end of the year. There will probably be different spectral masks in different parts of the world (just as there are different operating frequencies for Wi-Fi), but WiMedia is capable of adapting to the different masks: "we will have a standard for radios adaptable to any geography."

Every new PC now comes equipped with USB 2.0, and a wide variety of peripherals now incorporate USB. As wireless USB goes into PCs--probably soon into every new PC--it seems inevitable that more and more devices will add the technology. With more than 2 billion wired USB connections, it seems likely that in a few years there will be a billion or more wireless USB connections.

The Chip Maker's View--An Interview with Alereon

To get another view of WiMedia UWB, we talked on the phone with Eric Broockman, CEO of Alereon, a fabless semiconductor maker that's one of the UWB pioneers. We asked him whether he thought wireless USB would do away with cables, and he said "cabled USB provides power to devices, wireless provides mobility" so he thought most new devices would support both for a long time to come.

We asked how much cost UWB adds to devices. He said it initially will add $15 to the bill of materials (BOM), but the price of chipsets will come down as volumes build. Moreover, the BOM cost includes both PHY and MAC, but over time only the physical layers of UWB--the RF and PHY layers--will remain in hardware. "The MAC layer will disappear into SoCs, and it's half of the cost."

Eric thinks the first markets will be PC adapters and wireless hubs, and embedded products will follow. He thinks the early market will be in late 2006, with "high end cell phones and others" following in 2007. He predicts we'll see UWB built into PCs for the Christmas 2007 market with 2008 as "the breakout year"; by 2009 we'll see UWB as a standard feature in a third to half of new PCs.

The "Single Radio"

Eric pointed out that by 2008 UWB will be used for more than just USB. The 1394 Trade Association (1394TA) started working with MBOA and WiMedia in 2004 to develop "wireless 1394" using WiMedia UWB radios. The Bluetooth SIG has selected WiMedia UWB to add "a high speed/high data rate option" to Bluetooth. WiMedia UWB will also be used for TCP/IP applications.

WiMedia UWB Platform --> Click for larger pictureWiMedia has a nice diagram that illustrates how four upper-level protocol stacks--USB, Bluetooth, 1394 and TCP/IP--will operate with a common WiMedia UWB radio platform. Each of the four stacks is interfaced to the common UWB radio through a Protocol Adaptation Layer (PAL). Since each of the stacks and PALs is just software on top of an existing UWB radio and software, once wireless USB is built in there's almost no added cost to add the three other stacks. Thus UWB is positioned to become the "single radio" used for all applications where devices are relatively close to each other.

If wireless USB succeeds in the market, over time it's likely that every PC--desktop and notebook--will have a UWB radio and all four protocol stacks. Individual devices will select the higher-level stacks most suitable for their applications -- USB for printers, cameras and MP3 players; Bluetooth for phones; 1394 for video players; and TCP/IP for near-in communications between PCs.

Alereon perspective on Bluetooth --> Click for larger picture

Alereon's presentation shows how some of these devices might communicate once UWB is widely deployed. PCs and peripherals will communicate with wireless USB. Advanced cellphones will use Bluetooth-UWB to communicate with each other and with PCs, while using traditional Bluetooth for low-speed devices such as headphones. Portable video devices could transfer video to projectors and TV sets using either USB or Bluetooth.

Might Security Be a Fly in the Ointment?

As we talked with the people who are deeply involved in developing and promoting UWB, we came away convinced that it will play a very important role in the future of home communications, clearing away the maze of different cables now required to interconnect electronic equipment. But we expressed concerns that UWB is coming to market without any practical home experience.

Most communications technologies go through an early phase of field release before everyone has agreed to standards and tested for interoperability. This happened with Ethernet, earlier versions of 802.11, USB, and many others. It was a messy process, and inevitably many issues appeared. Only after this early field shakedown did we see agreement on finished standards, certification testing, interoperability plugfests, and approved logos.

By contrast, there has been no field testing of UWB. We will very soon see millions--and maybe then billions--of units in the field. Given the rigor USB-IF is bringing to the process, it's very likely the products will be interoperable. Our concern is how well they will work in the real world.

Perhaps our biggest concern relates to security. The "single radio" approach is extremely appealing--but fraught with risk if there turn out to be security gaps similar to those already encountered with 802.11 and Bluetooth.

The security model used in wireless USB is built on that used with the current cabled USB. The wire plays two key roles in providing security: the spec says it "connects the nodes the owner/user specifically wants connected" and "protects all data in transit from casual observation or malicious modification...". The wired USB specs do not include any level of security beyond the inherent security of the wire. "The goal of USB Security is to provide this same level of user-confidence for wirelessly connected USB devices."

We discussed this with a security expert who raised several issues after reading the specifications. He observed that the security mechanisms are built on the experience gained from Wi-Fi and Bluetooth. His major concern is that these mechanisms appear most suitable for short-term communications between devices; they're not designed to set up a channel that may last for months, such as the USB connection between a PC and a printer. There does not appear to be a requirement for periodic rekeying, so someone who wants to crack a key could see the same key in use for a long time. If someone did crack a key, there's almost no limit to the damage they could wreak, especially if keyboards and mice adopt wireless USB for connection to PCs.

We certainly hope these concerns prove unfounded. We expect to test these wireless USB devices later this year and will watch to see if the advocates are right in their predictions of widespread success.

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )