Ultra wideband (UWB) is not a technology it is a set of government issued regulations that allow well regulated access to a wide swath of RF spectrum.

The WiMedia Alliance specifies an interoperable technology for exploiting this spectrum within the regulatory rules, but how WiMedia devices are built has been left up to individual implementers. Decisions as to which technologies to deploy vary widely from company to company.

Technical decisions on how to actually build a WiMedia compliant device are largely dictated by the market or more precisely how different companies perceive the market. The first proposed application for ultra wideband was for moving video between a set top box and a TV screen.

UWB was proposed as a physical layer for IEEE 802.15.3 as this standard was well into development before the UWB spectrum was opened up by regulators. IEEE 802.15.3 was aimed at building wireless connections primarily for set top box to TV screen. Manufacturers of battery-powered portable devices also became interested in UWB, and by the time the WiMedia Alliance was formed, the MAC layer evolved to support portable ‘peer to peer’ connections as well as the set top box application. While the WiMedia radio has its roots in IEEE 802.15.3, it evolved to take on a much wider set of applications.

This split has led to a wide range of implementations of the WiMedia standard. One vendor might use multiple antennas and a SiGe front end to give a set top box a longer range and higher bit rate, and another vendor might build a single chip CMOS implementation in a very small package that can fit into a mobile phone. The set top box and mobile phone may have little in common in how they are implemented, but they will be able to recognise each other and at a minimum communicate by beaconing to share the frequency spectrum.

The potential market for WiMedia silicon is much larger than any single digital radio has been in the past. It is capable of distributing digital video around a room. It can be small and low powered enough to fit in portable media players and mobile phones. It is simple enough to replace the USB cable used with a digital still camera.

The market for replacing cables for PC desktop peripherals will happen as soon as costs are competitive with cables. The cost of entry into the mobile phone market is greater than USB cable replacement, but the volumes can be substantial. Any of these markets is large, but the combination makes for an unprecedented opportunity.

Many companies are chasing the rainbow looking for the pot of gold, and all are taking different approaches to implementation.

Staying Single or Not?
The major decision an implementer of WiMedia must make is whether to build a single chip device or a multiple chip device. Traditionally networking devices are divided into MAC and PHY for Media Access (how data is framed) and PHYsical layer. The WiMedia Alliance has defined a MAC / PHY interface which presupposes that that there will be separate chips for this functionality. If one does build a single chip implementation, it must be able to act as a PHY as to date interoperability testing has been done at the PHY level. The split between MAC and PHY as separate devices is an artifact of WiMedia’s history when ultrawideband was seen as an alternative PHY to an existing IEEE 802.15.3 MAC. 

WiMedia uses OFDM as its modulation technique. A description of OFDM is beyond the scope of this article. From an implementation point of view, it is a well understood but computationally intensive process that requires a large number of digital gates. WiMedia also requires mandatory AES encryption. The encryption can be considered either as part of the MAC or baseband depending on implementation.

Today most commonly the RF or radio transceiver function is built in SiGe though there are some companies that agressively sought out a CMOS implementation. A practical single-chip solution is where both RF and digital components are in CMOS, and reap the benefits of lowest cost and smallest form-factor.

Some RF implementations have chosen to exploit multiple antennas. The number of external RF components varies by implementation. Some vendors do the RF matching with passive components on a circuit board, and others embed the passive elements in a low temperature co-fired ceramic (LTCC) substrate. The resulting package is called a SiP or system-in-package. 
Similarly there are many variations in Host I/O interfaces. PCI, PCIe, SDIO, USB host, USB device and proprietary interfaces can all be found in various implementations. Each of these physical I/O interface standards require that the MAC speak protocols associated with these interfaces.

WiMedia radios are capable of running several different over-the-air protocols simultaneously. Certified Wireless USB, WiMedia’s WiNet and Bluetooth 3.0 are protocols either well defined or in the process of definition. In addition, some applications will use proprietary protocols embedded in standard WiMedia frame structure. Some have chosen to implement the
WiMedia MAC with a general purpose processor and multiple I/O interfaces, and others have chosen to optimise the MAC for a faster time-to-market Certified Wireless USB implementation.

While WiMedia radios are capable of running multiple protocols, they don’t have to. It would be possible to build a device that could be used as a wireless mouse. It would have to be very inexpensive and probably would be incapable of acting in any other role than that of a wireless mouse or keyboard. In order to meet the cost requirements for a wireless mouse, the solution would have to  be single chip.

Packaging: Size Matters
Experience with Bluetooth and WiFi suggest the least expensive solution that can meet the specification and satisfy user expectations will win the majority of the market. There may be a niche market for high-performance, but for the most part, price is the major factor.

Mobile devices such as phones and portable media players demand very low power consumption and very small packages, and price remains incredibly important.

In the shorter term the first devices to ship will be devices for desktop and notebook PCs and powered Certified Wireless USB hubs. The early devices can afford to burn a bit more power and can be a bit larger. There is plenty of room for passive components in a wireless USB hub.

The digital still camera lies somewhere in between. Size is important, but cameras only turn on the radio when sending pictures to another camera, a PC or a printer. Digital still cameras are expected to be early adopters of WiMedia-based solutions.
The driving force behind early adoption into mobile devices are solutions implemented as ad-on cards, such as Staccato Communications’ SDIO card (SC3226R SDIO for Native Certified Wireless USB) which utilises a single-chip CMOS SiP packaged solution.

The trend is to smaller cheaper packaging. Ultimately low-cost and low-height packaging will utilize standard IC packaging technology. Examples are:

• Quad Flatpack No-lead (QFN), or Micro Lead Frame (MLF)
• Ball Grid Array (BGA)
• Land Grid Array (LGA)
• Wafer Chip Scale Packaging (W-CSP)

Today SiPs are based on LTCC and will move to stacked die in the not too distant future.

The details of how these packaging technologies work are not so relevant. The important issue is that packaging technology is advancing just as Moore’s law is lowering the cost of CMOS chip production. The mobile phone industry is demanding smaller and smaller packages and better and better RF performance along with miniscule power consumption.

For the foreseeable future Moore’s Law and evolving packaging technology are conspiring to make a rosy future for WiMedia. Moore’s Law allows more complex digital devices, but as transistors get smaller it becomes possible to support higher frequencies with CMOS implementations. As more functionality is pressed into single chip packages, automated testing becomes easier as does automated assembly. As the virtuous circle turns, we expect to see WiMedia evolve to better performance at lower prices and into more and more product.

While there may be some market acceptance today for first generation WiMedia products based on multiple chip solutions, the inexorable trend will be single-chip, low-power, low-cost solutions in small RF efficient packages and entirely in CMOS.

Billy Brackenridge, product system architect at Staccato Communications