Published in 2003 IT/AV Report

Handicapping the Wireless AV Horserace
By By Neal Weinstock

      Wireless LANs are the epicenter of industry heat. The market research firm In-Stat reported 8 million WLAN units sold in 2002, and expects 33 million in 2003. That huge increase may be as much about manufacturers’ wishful thinking as consumer demand. Nevertheless, wireless chips should continue on a rapid downward cost curve. The IEEE standards 802.11b, 802.11a and 802.11g have been approved and widely adopted, with more iterations coming. Bigger-bandwidth versions of IEEE 802.15 (Bluetooth) and Ultra Wide Band (UWB) are on the way. But can these standards be put to use for AV?

‘They Must’
      They must, according to Duncan Kitchin, wireless LAN architect for Intel. “802.11b in home networks was driven by broadband internet connection sharing,” he stated. “802.11a in home networks will be driven by high bandwidth multimedia streams between devices in the home.”
Wireless connections would mean greatly faster and easier AV installations. Wireless video cameras for security, process control and other industrial uses can be a tremendous productivity enhancement. For example, in many cities, people are in the habit of checking traffic webcams before deciding on their commutation route. Wired webcams involve a not inconsiderable installation expense. See http://nyc.gov/html/dot/html/motorist/atis.html, the website of the New York City Department of Transportation Real-Time Traffic Camera advisory. There are just 82 digital still cameras mounted around New York, and only 17 streaming video cameras. That’s not much over thousands of miles of roadways.

You Might Never Know
      With miles between cameras aimed at major routes such as the Brooklyn-Queens Expressway, you’d never really know whether a webcam’s clear-looking traffic means your drive to the airport will be smooth, or that there’s a jam-up just before the camera’s location. If a wireless connection can reduce overall installation and maintenance costs enough to put up several times the number of webcams, the benefits will strongly outweigh the occasional environmental or contention-based interference causing signal outages.
      Wireless before the days of 802.11 would have meant licensed spectrum, at from 12 to 26GHz depending on the location. Licenses in these frequencies are costly, and may not be available in a particular area. Higher-bandwidth systems are also far more power-hungry than 802.11. Still-image digital cameras and 802.11 may even be powered by the solar cells that are now ubiquitous on emergency phones at roadsides.
      Although audio takes up far less bandwidth than video, wireless may happen more slowly in audio installations, because latency must be tighter in audio. Glitches in surveillance webcams are OK, but quality video can’t have latency greater than 25ms. Audio will sound murky if latency between stereo pairs or other closely aligned monitors comes in greater than about a tenth of that, or 2.5ms (though “golden ears” will insist on much less). Voice-quality audio, such as the payload in an airport PA system, requires about 20ms latency at maximum. According to Intersil, the largest supplier of 802.11 chips, 802.11b at 11Mbps can carry up to 10 isochronous voice channels (at typical voice quality of 44kHz) by using the standard’s PCF option, with 4 to 6ms latency/channel. After many years of work on PCF over 802.11b and improved audio compression codecs, various schemes allow multiple streams of quality audio over WLAN.

On the Surface
      On the surface, carriage of AV over IEEE 802.11a and 802.11g would seem to be a no-brainer. Both offer 54Mbps, with effective carriage capacity of a 20 to 30Mbps payload. In comparison, DVD video streams are 6Mbps; theoretically, four or five could be accommodated. So could a couple of dozen 96k digital audio channels. But there are limits to the attractiveness of these standards for real-time media.
      According to Intel’s Kitchin, both the established 802.11a and b standards’ MACs are only about “50% efficient.” That’s due not only to the collisions, backoff time and acknowledgements inherent not in Ethernet, but to additional features that were built into the wireless standard to enhance its ability to deal with new devices introducing themselves to the cloud.
Other 802.11 devices roaming into the cloud might insist on network access; a security scheme that denies them access is easy to deploy, but takes up some signal headroom and may limit the network’s usefulness for AV apps. Any other RF emitters nearby may interfere with connections. Examples include other 802.11 devices, microwave ovens (mainly affecting 802.11a at 5GHz), 2.4GHz portable phones and Bluetooth (affecting 802.11b and g at 2.4GHz).
      Interference is a bigger problem in the more populated 2.4GHz spectrum used by 802.11b and g, a lesser one in the more recently opened 5GHz range used by 802.11a. But some of the problem at lower frequencies is a facet of the greater reach they make possible at any given power. IEEE 802.11a only has about a 50-foot range; signals in the 2.4GHz range can be trusted for about three to four times that distance. Less range may be inconvenient, but it also means that the microwave oven in the commissary down the hall can’t reach your signal.
      Finally, different environmental objects may scramble any radio signal, depending on the characteristics of the frequency band. Steel is a big problem for the 5GHz range, so 802.11a may not work in a high-rise office building…or it may only work within a single room there. That’s all you might expect anyway from 802.11a’s 50-foot distance. Masses of concrete can also pose problems. All of which tends to limit best performance by these networks to the setting of a home or a small suburban office or fast-food restaurant.

Enhance 802.11a
      Kitchin chairs the 802.11e task group, which expects to enhance the 802.11a standard by allowing delivery of real-time traffic with the tightest possible latency. This will imply many choices among a great number of possible new ways of prioritizing certain traffic and denying access to other traffic. So, in a year or two, we will probably get a new 802.11 standard that allows QoS for multiple streams of real-time AV.

Newer Protocols Spur Action
      Major spurs to action on improvements in 802.11 are the ongoing improvements in competing protocols. The best known is often called wideband Bluetooth, or IEEE 802.15.3. The most radical is Ultra Wide Band. Both offer opportunity to step back, rethink bandwidth use and QoS for AV, and produce standards that are better than 802.11. After all, 802.11 includes an accretion of iterations dating back decades.
      802.15.3 is by far the less radical of the new standards, and much more directly competitive with 802.11. It aims for up to 55Mbps as far as 70 meters, and peer-to-peer connectivity with QoS for AV. It also supports low-power devices and 128-bit encryption.
According to John Barr of Motorola, chairman of the IEEE 802.15.3 Task Group, “Compared to 802.11 at 2.4GHz, an 802.15.3 2.4GHz PHY system causes less interference because it occupies a smaller bandwidth and transmits at lower power levels.”
      If the peer-to-peer and AV-oriented nature of 802.15.3 sound much like the benefits of Firewire, that’s intentional. The standard was inspired by 1394 and can also be used, according to Barr, as a wireless 1394 replacement, up to that 55Mbps bandwidth limit. This would not be the first wireless protocol based on 1394; in Europe, HiperLAN is also known as “wireless 1394.”
As with 1394, what’s important about 802.15.3 is on a level above the physical. The 802.15.3 PHY isn’t much different from 802.11’s. So it should come as no surprise that the proponents of 802.15.3 are looking at alternate PHYs, especially UWB.

UWB
      All other radios today use limited bandwidth, and modulate frequency or amplitude within that band. Instead, UWB sends patterns of extremely short (sub-nanosecond) pulses across vast areas of spectrum. With frequency ranging from hundreds of thousands to billions of pulses per second, and using pulse-position modulation, binary keying and other techniques, UWB achieves strong signals even at low power. It looks like white noise to an existing radio receiver, and coexists with it.
      The technology was developed for the US Department of Defense (DoD), and remained classified until a few years ago. Recently, the Federal Communications Commission persuaded the DoD to go public with UWB. Commercial use at low power is now conditionally allowed.
Various spectrum licensees are lobbying against UWB, saying they fear it may affect their signals. The military, however, has long been using UWB without affecting those signals. Manufacturers of existing 802.11 can’t be happy about UWB, either; it is a disruptive technology that may upset their apple cart.
      The FCC has limited UWB power spectral density (PSD) to -41.25dBm/MHz. The PSD limits for the 2.4GHz ISM and 5GHz U-NII bands used by 802.11 and current 802.15.3 are more than 40dB higher per MHz. Many observers believe that this highly conservative limit on UWB power will be raised eventually, because interference with existing communications is a non-issue. Even at such low power, UWB is capable of carrying 100Mbps or even higher data rates. UWB is, however, currently limited to short distances.
      The leading company in commercializing UWB, a Virginia start-up backed by Motorola, Texas Instruments and Cisco Systems called Xtreme Spectrum, already has a PHY chip that does 25Mbps at 50 meters. They and the standards working groups have goals of specifying 110Mbps service at 10 meters distance, and 400Mbps at five meters.
      That 400Mbps at five meters high-capacity/short distance rating, if combined with the 802.15.3 standard in a MAC or link layer sitting on top of UWB, would coincidentally yield almost the exact characteristics of IEEE 1394a, without wires. This new hybrid standard is now known as 802.15.3a.

The Network Effect
      Which brings us back to wireless 1394. HiperLAN was not widely adopt- ed because 802.11 edged it in the market with lower pricing and wider use. Networking technologies are among the best examples of the “network effect”; users want to use what other users are using, so they can connect to more users. As soon as one among any number of competing technologies seems to be the more widely used, almost all potential users adopt it.
      Could 802.11 already have such a head start over 802.15.3a that the game is over before it starts? Probably not this time. There is a confusing profusion of 802.11 standards, and they are mostly mutually incompatible. Manufacturers will have to move rapidly to provide systems that connect at both 2.4 and 5GHz, and that easily upgrade to whatever becomes most popular as a future QoS standard: 802.11e, or g, or maybe something else.
      Any consumers stuck with 802.11b and a lot of interference, or 802.11a where most of their friends use 802.11b, are likely candidates to switch to 802.15.3a when they upgrade. Electronics manufacturers that want to build in high-bandwidth, peer-to-peer connections to link a camera or loudspeaker will simply choose the best technology.

UWB’s Hurdles
      UWB does, however, face other significant hurdles. For now, it’s allowed only in the United States. UWB requires clear proof that it does not interfere with existing radio at greater power ratings, and it requires new hardware. 802.11 may yet pull out another victory.
      AV installations win either way, though. As the hardware companies chase their grail of consumer demand for wireless home AV networking, they will bring greater labor savings and quicker, safer installs to all networked AV.