<font color=orange>Roads diverge to wireless net service quality
</font>By Rick Merritt -- EE Times -- June 2, 2003 (3:19 p.m. EST)
SAN MATEO, Calif. — Engineers hammering out a quality-of-service (QoS) standard for 802.11 networks have agreed to disagree by writing two distinctly different mechanisms into a draft standard that could be ratified early next year. Computer and consumer electronics companies are working to ensure their competing approaches will not raise interoperability issues, although they fully expect the approaches will have to battle it out in the marketplace.
Quality-of-service is a key technology enabling the wireless delivery of voice, music and video. Some PC makers had hoped a high-bandwidth version of 802.11 with QoS could be a unifying home-networking technology for both PC and consumer gear. But many consumer electronics manufacturers say that while the QoS work is key to their adoption of 802.11 in products that could start to ship this year, no single wired or wireless network will serve all of their needs.
Computer industry companies such as Intel Corp. and Microsoft Corp. have rallied support for Enhanced Distributed Channel Access (EDCA), a best-efforts mechanism in which devices negotiate for bandwidth based on up to eight levels of priority using traditional 802.11 probabilistic methods. A group consisting of Atheros Communications Inc., Cisco Systems Inc., Intersil Corp. and others has already defined a subset of the pre-standard EDCA features, called Wireless Multimedia Enhancements, and implemented it in products.
Consumer OEMs and some carriers argue that the scheme does not provide the bandwidth and latency guarantees they need for glitch-free voice, music and video. They are lobbying for a polled-access scheme, broadly known as the Hybrid Coordination Function, in which access points regularly poll and grant devices exclusive access to an 802.11 channel in a more deterministic fashion.
EDCA backers say the HCF approach is at odds with quality-of-service techniques adopted on the Internet. And they say the deterministic methods in HCF are more likely to lead to packet collisions with neighboring 802.11 networks.
HCF backers counter that consumers may think their snazzy digital phones, stereos and TVs are malfunctioning if they hear pops in audio or see jerks in video because of shifting demands on an EDCF-based home net or a public Wi-Fi hotspot.
Some carriers argue they need HCF to deliver data or voice reliably over a paid-for public hotspot. EDCA backers accuse the carriers of using HCF in a "spectrum grab" of unlicensed Wi-Fi bands.
"Some of us don't think that's the right idea. There's a lot of money at stake here, and people are playing games," said one .11e participant, who asked not to be identified.
Engineers say the heated religious debates have settled down in recent months, with developers converging on the goal of getting out a standard that mandates EDCA but allows for an HCF option. That could pave the way for both approaches, letting the market sort out what it wants.
Both camps are hammering out the details of their approaches as they refine the .11e standard. Chip makers say either version can be implemented in firmware, without specific hardware support.
The current draft, 4.2, in the .11e working group states that any 802.11 device should be able to recognize and respond to an HCF device but does not have to share traffic with that station. Because the standard is not expected to be ratified until next year, engineers say it's still possible HCF could become a mandatory part of the spec. But that becomes less likely as time goes on.
The Wi-Fi Alliance, a broad ad hoc 802.11 industry group, expects to complete a testing and branding program in about two months for the .11e QoS standard. A task group in the alliance is still debating whether testing and branding will be done by feature, such as voice and audio/video distribution, or by QoS approach.
The test procedures will be key for HCF because it requires an admission control scheduler that's beyond the scope of what will be defined in 802.11e.
Just how all these details settle out could be key to determining how broadly 802.11 spreads from use in computers today to consumer gear in the future.
"As time goes on, QoS will be extremely important for the broader adoption of Wi-Fi, so it's very important we address this now," said Dennis Eaton, chairman of the Wi-Fi Alliance.
Intersil believes "802.11 will find its way into the consumer electronics space," said Jim Zyren, strategic marketing manager for wireless at Intersil (Milipitas, Calif.). "The question is how far and how fast."
"Pretty much all the Japanese consumer companies I can think of-including Matsushita and Philips-are in agreement over HCF," said John Kowalski, manager of communications systems at Sharp's offices in Camus, Wash., and a participant in both the .11e and Wi-Fi QoS groups.
Sharp implemented an early version of HCF for a wireless LCD now shipping in Japan based on an 802.11b PHY and a proprietary MAC chip. But, going forward, 802.11 will be just one of many home net technologies Sharp will employ.
"We don't want to put all our eggs in any one basket. Since we want to sell end equipment, we have to take a broad view [of networking]," said Kowalski.
Scott Smyers, vice president of Sony Corp.'s Interconnect Architecture Lab in San Jose, Calif., takes a similar view, noting that any wireless technology has inherent limitations. "It's not like there's one skunk works [home networking] project that will come up with a standard that will swoop down on this problem and solve everything," he said.
Subset of PC camp's scheme sets four priority levels based on 802.11.
Smyers said he does not foresee Sony's taking any wireless home networking products to market until the relevant standards are ratified. "We need QoS in wireless in a way that represents an industry consensus, but the jury is still out over how the industry will resolve the issue," he said.
With a chip maker's perspective, Zyren said, "Intersil feels EDCF is good enough, but if CE companies feel strongly HCF is what they want, we will respond to that."
"There are some customers who want [HCF] and a small subset of those customers who actually need it," said Greg Chesson, director of protocol engineering at Atheros Communications (Sunnyvale, Calif.). "We will almost certainly implement it." But he also waxed philosophical: "Everybody and their dog is looking at techniques for better voice and video on a wireless network. Who knows if anyone is going to buy this stuff? It's a shot in the dark at the next big thing."
Many consumer electronics companies have longstanding investments in wired 1394 networks as the best route to QoS between devices. But the 1394 Trade Association now sponsors work that puts the 1394 protocols over both 802.11 and 802.15.3 wireless networks.
Startup Appairant Technologies Inc. is designing a MAC for both the 802.15.3 wireless standard, which does not require an access point, and .3a, the emerging ultrawideband (UWB) standard. "Even if you use a gigahertz radio, the most you will get from 802.11 is 72 Mbits/second because you have the bottleneck of pushing everything through the access points," said Tom McGovern, chief executive of Appairant (Attleboro, Mass.).
The UWB work will take 18 to 24 months to reach a standard, has no defined PHY as yet and has no regulatory approval outside the United States, he noted.
Thomson Inc. is taking a unique approach, with plans to roll out a merchant chip set this year that handles both 802.11a and HiperLAN 2 at 5 GHz. The company expects at least two others will roll HiperLAN 2 chip sets this year using the deterministic quality-of-service capabilities of HiperLAN to handle video.
Thomson demonstrated an early version of the chip set at the Consumer Electronics Show in January that transmitted video over the 5-GHz link. Thomson is expected to field systems-such as plasma displays and multimedia servers-using the chip set at next year's CES. "The whole [consumer] market is going to [networking], and we have to have a solution," said Bill Mengel, director of emerging technologies.
Mengel said he is tracking UWB but does no active development on it. He is, however, considering a next-generation chip set that would use both 802.11g and HiperLAN 2 across 2.4-GHz and 5-GHz links.
"Wireless is hot, and at some point we are going to find the right recipe," said Chesson of Atheros. "I don't think anyone knows what it is yet, but there's a lot of experimentation going on."
commsdesign.com |
