Ultrawideband and WiMedia

Ultrawideband and WiMedia
By John Canosa, Embedded Systems Programming
March 20, 2003 (12:14 p.m. EST)
URL: http://www.eetimes.com/story/OEG20030319S0019

Short-range wireless communications via emerging ultrawideband techniques are hot. Here's a look at the latest developments, as they'll be presented at the Embedded Systems Conference later this month.

In a crisp, clear day in 1901, Guglielmo Marconi used a makeshift combination of coils, capacitors, and wires to create spark-gap transmissions that broadcast Morse code sequences to the other side of the Atlantic ocean. Hours later he received a return message from the widow of a Nigerian general with an urgent business proposition for him. That last part didn't actually happen, but with the successful wireless transmission of data at approximately 10bps, Marconi did set off a wireless revolution that is still in progress today.

Some time after Marconi's revolutionary transmission, other scientists began wondering how multiple transmitters and receivers could share the same airwaves w ithout interfering with each other. The concept of separation by frequency was born and has been with us ever since.

To better understand separation by frequency we need to reawaken some long dormant brain cells. Remember the correlation between the time domain and the frequency domain as expressed by the Fourier transform? Marconi's spark gap created pulses of energy that were extremely short in duration. Taken to the mathematical extreme, these pulses represent an impulse function that has an infinite amplitude and occurs for an infinitesimal time. As demonstrated in Figure 1, a pulse of this type literally radiates as every frequency in the spectrum.

Figure 1: Fourier transform of an impulse function

Since the transmission occurs in many frequency bands, it becomes difficult to filter out information or noise that is not intended for the receiver. The concept of separation by frequency, o r narrowband transmissions, uses a specific frequency, which is easily isolated from other frequencies via filtering, to transmit information. As shown in Figure 2, the Fourier transform of a sine wave appears as a very narrow pulse in the frequency domain, making it easy to filter out other transmissions.

Figure 2: Separation by frequency

To keep people from using all the spectrum at will and interfering with valuable wireless services such as emergency, military, and TV broadcasts, the frequency spectrum has been divided up and regulated by various governments. In the United States, spectrum allocation is the domain of the Federal Communications Commission (FCC).

The downside of narrowband transmission is that the limited bandwidth of the signal eventually places limitations on the amount of information the signal can transmit. And with the frequency spectrum becoming ever more crowded wh ile data transmission needs are increasing, new techniques have been developed that attempt to provide high-speed data transmission without using up and interfering with other reserved parts of the spectrum. One new technique is really a very old technique of sending information in very narrow pulses of data, just as Marconi did over a hundred years ago.

Ultrawideband
Ultrawideband (UWB) transmitters use a series of very short pulses to communicate with other devices. The beauty of this approach from the transmitter side is its simplicity. Complex filters, mixers, and other such RF paraphenalia are unnecessary. The pulse is also transmitted at very low power levels-low enough, in fact, and by regulation, to be below what the FCC defines as the noise floor for narrowband systems.

While low power levels protect existing systems from interference, they complicate the task of an ultrawideband receiver. Not only must the receiver be able to pick up a signal out of what is normally considered noise, it must also deal with the legacy systems that transmit data at much higher power levels within their specific narrow bandwidth limitations.

Ideally, UWB would use as much of the frequency spectrum as possible to transmit data as fast as possible. However this usage would interfere with other users of the spectrum. The developers of UWB recognized this issue and have complied with the FCC Part 15 regulations that limit the power output of any radiator to certain levels. Figure 3 shows what location Part 15 regulations limit. Unfortunately, low-power levels raise the question of how to determine what is noise and what is information. This is where the UWB vendors begin to differ in their approach. Since we're talking about pulses in the time domain, only a couple of parameters can be varied to transmit information. Specifically, we can control either the amplitude of the pulse or the spacing between pulses. These techniques have the fancy names of pulse amplitude modulation (PAM) and pulse position modulation (PPM).

Figure 3: Ultrawideband in the radio spectrum

All of the modulation methods place ultrashort (sub 1ns) pulses into a pulse train with repetition rates upwards of 1GHz. Information, in the form of ones and zeros, is encoded into the pulse stream by varying the position of the pulses within a fixed window (PPM), by on/off keying (OOK) encoded ones and zeros with the presense or absence of a pulse in a window, or by binary phase shift keying (BPSK) ones or zeros by altering the phase of the pulse (does it start out going positive or negative?).

While claims of low power and simple circuitry for UWB implementations have some merit on the transmit side, the receive section must be capable of separating pulses from other noise sources and dealing with multiple reflected versions of the same pulse. This tends to make receiver implementations complex, requiring significant processing capability and eliminating potential power-consumption savings on that side.

Another area of concern is the very issue that led to the adoption of narrowband techniques in the first place. That is, how can different UWB systems coexist in the same environment without interfering with each other? Several aspects to this problem include:

How does a receiver determine what pulse came from what transmitter?

How can we prevent several transmitters from additively generating more total noise than is allowable under FCC regulations?

How can we prevent a nearby transmitter from completely masking the transmission of a device that is located farther from the receiver?

Techniques borrowed from the narrowband world are being discussed to solve some of these problems. For example, time division multiple access (TDMA) systems give each device in a network individual time slots in which to transmit and code division multiple acce ss (CDMA) systems assign specific codes to each device. Techniques like these require higher level protocols and network controllers to coordinate usage, all of which require extensive standards work to ensure interoperability among devices from different vendors.

WiMedia
UWB proponents focus on its potential to achieve 500Mbps of throughput at ranges up to 100 meters. Such capabilities would put UWB in the realm of wireless LANs. However, more realistic near-term expectations of 100Mbps at 10 meters and reasonably low-power consumption put UWB squarely in the realm of wireless personal area networks (WPAN).

The focus in the WPAN space is on a new standard called WiMedia (www.wimedia.org). WiMedia is most often used to create ad-hoc, interoperable networks for audio, video, and computing equipment such as camcorders, digital cameras, speakers, and home theater systems.

WiMedia defines a series of application profiles (such as a dig ital imaging device or a video recorder) that sit on top of the transport, control, and service discovery protocols. These protocols in turn sit on top of the IEEE 802.15.3 Media Access and Physical layer definitions, as shown in Figure 5. The 802.15.3 MAC is important because it attempts to provide ease of use, quality of service, and a security framework for a high data-rate WPAN.

Figure 4: Ultrawideband modulation techniques

Figure 5: WiMedia protocol stack layers

Quality of service is enabled by using a TDMA-based system that's managed by a network controller node. When a device requests a certain amount of bandwidth, the host controller grants specific time slots during which the device may transmit data. The standard also addresses security by incorporating a security framework th at currently supports three public key encryption techniques. The reasoning behind specifying a security framework rather than a single security implementation is to avoid the minor disaster suffered by 802.11b when WEP was designed into the specification and later found vulnerable to attack.

Currently, UWB is proposed only as an alternate physical layer for WiMedia. The official supported physical layer is the orthogonal frequency division multiplexing (OFDM) narrowband system used in IEEE 802.11g. OFDM offers 22Mbps at 70 meters and 55Mbps at 10 meters, so UWB offers a bandwidth upgrade. The caution here is that multiple physical layers can lead to interoperability problems; unless competing UWB vendors can agree on a modulation scheme quickly, the well-defined 802.11g narrowband implementation may win.

Aside from basic data transmission, the time-domain nature of UWB adds some intriguing application possibilities. One possibility is ranging and location. Using triangulation techniques, UWB can pro vide highly accurate location information. The ranging capabilities are endless, with a lot of attention being placed on collision avoidance systems in automobiles. And finally, the pulsed nature of UWB makes it useful as a radar-based imaging system. Already on the market are UWB radar-based imaging products that enable rescue workers to search for survivors in the rubble of collapsed buildings.

Interfacing
One of the more welcome trends over the years is that chip vendors are realizing that even the best communications components are useless if difficult to use. This is particularly true in the area of software interfacing. The WiMedia organization and the drivers of the 802.15.3 specification have defined a frame convergence layer that sits between the MAC and the upper layer protocols. The Frame Convergence Layer is intended to provide an abstraction of the MAC layer that maps to one of several well-known interfaces such a PCI, USB, Fir ewire, Ethernet, and simple memory mapped I/O models. UWB implementations currently come in the form of chipsets, with as many as four ICs required for a complete solution. Next generation devices will likely pare this down to two chips-an RF chip and a baseband processor. A typical implementation is shown in Figure 6.

Figure 6: Typical UWB chipset

It should be noted that a very gray line demarks what parts of the WiMedia stack are implemented in hardware versus software. Profiles and security infrastructure are likely to be implemented in software for the foreseeable future and will be the area on which embedded developers focus. Wireless nirvana?

Once again we face a new technology that is being hyped as the Next Big Thing. As the hype cycle sets expectations higher and higher, the disillusionment when reality sets in will only be deeper and deeper. Just ask your local Bluetooth vendor. U WB is an exciting new technology with some very impressive capabilities. But claims of 500Mbps at 100 meter ranges for $5 need to be looked at with a healthy dose of skepticism. Realistic expectations over the next year or so are closer to 100 to 200Mbps over 10 meters, with a bill of materials approaching $30.

For applications such as positioning and radar imaging, UWB opens up a whole new world and cost structure for some products. It could really lead to some innovative, new products. From a data transmission perspective, UWB is currently only an intriguing possibility if you a designing a closed system that does not depend on multivendor interoperability. For those of you looking to go to market soon with WiMedia devices, it might be best to stick with the currently approved OFDM narrowband systems and look to UWB next generation products.

It behooves the proponents of UWB to settle on some common modulation schemes and to close ranks around an interoperable model. For the latest information abo ut standards, visit www.uwb.org.

John Canosa is the chief scientist at Questra and a contributing editor of Embedded Systems Programming. He earned his BSEE from Clarkson University and MSEE from Rochester Institiute of Technology. Contact him at jcanosa@questra.com.