re: wireless-LANs
<< the question of auctioning licenses (never a fan, but thought they must know what they were doing) >>
Americas gift to Europe. Europe was scratching their heads and wondering what we were doing when we started auctioning off wireless spectrum in 1994. A&B Block went great ... then came C Block. YUK! Financing dried up (sound familiar). Still not resolved. It will all work out in the end. Maybe beauty contests are better. Don't know. We auction off little chunks of geography, and little chunks of spectrum. If we ever auctioned nationwide with no spectrum cap and a big chunk of spectrum ... watch out. UK, Germany and Italy prices would seem VERY paltry. Look what 15 MHZ of NYC just brought in spectrum that is still contested.
<< The vision has always been PAN, LAN and WAN together >>
Standards and spectrum are issues but this is promising ...
>> FCC Decision Opens Door To High-Rate Wireless LANs
Patrick Mannion
EE Times
03/09/01
The Federal Communications Commission may change its definition of spread spectrum to enable higher-rate wireless LANs, one of the agency's top officers said this week.
"The changes will add more flexibility to the rules to accommodate some of the recent advances in technology," said Julius Knapp, chief of the policy and rules office at the FCC's Office of Engineering and Technology. Knapp will announce the FCC's decision to revisit the rules - but stop short of saying just what the changes will be - at the IEEE Working Group meeting in Hilton Head, S.C., next week.
Prompted by pressure from wireless-LAN proponents, the potential changes end months of speculation about the legality of three possible paths to 22-Mbit/ second wireless LAN radios in the 2.45-GHz band. Those proposals, now before the IEEE 802.11's Task Group G, all meet the spectral and power-emission requirements for spread-spectrum radios, but don't align with the actual definition.
The proposals come from Texas Instruments Inc., Intersil Corp. and Supergold Communications Ltd., a three-year-old startup based in Dublin, Ireland, which arose out of work at the University of Leeds in the United Kingdom.
All three will come under scrutiny in Hilton Head, and by Friday, "We should have it narrowed down to two," said 802.11 Working Group vice chair Al Petrick. Which two remains the subject of much debate, as advocates spar over issues such as ease of implementation, cost, power and patent rights in using orthogonal frequency-division multiplexing as an access method.
Knapp said the FCC will let IEEE decide which coding proposal best serves the purpose, and will work with IEEE to modify spread-spectrum rules to let the selected proposal operate.
Doubling the 11-Mbit/s data rate will extend the life of the large installed base of IEEE-802.11b wireless LANs, and the upcoming addition of security and quality-of-service features will open the door to fully enabled networks delivering streaming audio and video and voice-over-Internet Protocol services. But the advent of these souped-up .11b LANs could steal the thunder from the next generation of WLAN, in the form of IEEE 802.11a and HiperLAN 2. Those schemes — both primed for year-end rollout - work in the cleaner 5-GHz band, but are generally seen as more costly to implement.
The high-rate .11b work is "too little, too late," said Rich Redelfs, president and chief executive officer of Atheros Communications (Sunnyvale, Calif.). "By the time it's out there, and allowed by the FCC, we'll have 54-Mbit/s 802.11a devices out - and in a cleaner band to boot."
Atheros plans to release its two-chip, all-CMOS .11a solution this summer for under $35.
"Some say that .11b's lifetime is limited," countered Chris Heegard, chief technical officer of TI's home and wireless networking group. "[But] I see both being popular: 801.11a won't replace 802.11b, or vice versa. The penetration of 802.11b into public places, such as the Dallas-Fort Worth airport, will keep it around for some time to come."
Craig Mathias, principal at research firm Farpoint Group (Ashland, Mass.), believes the .11b enhancements are "worthwhile doing, as an extension to what's there." But in Mathias' view, the 5-GHz WLAN standards have some advantages that .11b cannot match. Even with high-rate .11b, he said, "You're still looking at less range, greater susceptibility to interference and all the other problems facing 5 GHz - but in a band that's not as clean."
Security, QoS
Also adding juice to .11b are new security and quality-of-service (QoS) features. At the Hilton Head meeting, a separate IEEE entity, Task Group E, will vote on a slew of proposals for QoS support and more robust security in the 802.11b media-access control layer (802.11a uses the same MAC).
Those efforts came into sharp relief recently when researchers at the University of California, Berkeley, said they had cracked the 40-bit Wired Equivalency Protocol now specified for .11b networking. The security weaknesses have long been known - hence the work by Task Group E (TGe) to add a 128-bit key and a longer initialization vector - 32 bytes instead of 3 bytes - to help mitigate the problem.
Higher data rates, better security and QoS are being folded into .11b just as 5-GHz wireless-LAN technologies such as 802.11a and HiperLAN 2 are converging on an end-of-year rollout schedule. HiperLAN 2 has long been touted for its QoS and security, but the work by the TGe group - scheduled to be finished by the end of the year - might well make these advantages moot. Given that .11b and .11a share a MAC, the work applies to HiperLAN 2's 5-GHz rival, 802.11a, as well.
This leaves HiperLAN 2 relying on its internetworking-convergence layer and additional power- and frequency-control features as its value-added proposition.
But HiperLAN 2's ATM genesis and its MAC-layer functionality are crucial differentiators, said Per Wiklund, product-marketing manager for Ericsson's business networks (Kista, Sweden), and are the reasons why Ericsson picked it for 5 GHz. The company recently demonstrated its HiperLAN 2 ASIC prototype solution in a PC-card client terminal and an access point.
"We were particularly attracted to the convergence layer and power-control features of HiperLAN 2," said Wiklund, referring to the layer within the specification that handles internetworking between HiperLAN 2 and local networks like wireless 1394 and wide-area networks such as third-generation (3G) wireless cellular schemes. "UMTS operators see this as a nice complement to their networks," he said, giving HiperLAN 2 a leg up against 802.11a.
Ericsson plans its first product release by early 2002. In the meantime, the company sees the rival 802.11a WLAN as taking the harder hit from competition with high-rate .11b.
The 1394 wireless group is working on a 1394-over-802.11a convergence layer, noted Duncan Kitchin, vice chair of the TGe effort. But Kitchin doesn't see the convergence layer as the proper place to do multinetwork roaming. "There's a general client mobility issue to be addressed," he said, "but it's not something that needs to be built into the .11a standard or the 3G standard. Both are able to transport IP, so there's no reason they can't coexist."
Companies such as Texas Instruments, which plays in both the .11b and .11a camps, foresee a transitional period during which both will serve their purpose. TI's high-rate, 22-Mbit/s .11b proposal derives from work done at an acquisition, Alantro Communications. Heegard, who hails from Alantro and worked on the .11b spec there, said the plan is based on a combination of two technologies: complementary code keying (CCK) and packet binary convolutional coding (PBCC).
CCK is the basis of 802.11b, but Alantro also slid PBCC into the specification as an option. "So, we have PBCC in the standard for both 5.5- and 11-Mbit/s .11b," said Heegard," as well as "a new version of that for 22 Mbits/s."
The complex coding in TI's high-rate radio proposal has the same spectral mask as CCK, operates in the same environment, but doubles the data rate. "As a result, we see our proposal as being much easier for the FCC to accept, as it looks exactly like the existing signal," Heegard said.
By contrast, Heegard said the orthogonal frequency-division multiplex (OFDM) proposal championed by Intersil looks quite different. "We've shown that if you introduce an OFDM signal at the same power as a CCK signal into an 802.11b environment, the OFDM signal is a more severe interferer as a result of the peak-to-average ratio," said Heegard. "As a result, the FCC should think carefully about letting OFDM into the 2.45-GHz band."
Heegard pointed to implementation issues with OFDM as well. "As a technology, OFDM is fine, and it can be made to work in the 2.45-GHz band," he said. "However, it does require a rethinking of the radio and faces patent issues with companies such as Wi-LAN, which has declared its intention to seek royalties for users of OFDM." TI is licensing its PBCC patents royalty-free.
Heegard also said that "Intersil is trying to shoehorn in a different modulation scheme in a backward-compatible manner, which brings in a lot of overhead that won't go away."
But according to Jim Zyren, marketing director at Intersil, the TI camp should be more concerned with its own scalability issues. "OFDM uses an efficient fast-Fourier transform for the channel estimation - it's very efficient and scales easily," he said. "Hence the complexity of the radio doesn't go up with the data rate. The equalizer in the PBCC implementation, on the other hand, doesn't scale linearly, but exponentially."
This was refuted by Heegard, who pointed to the linear-scaling capabilities of TI's practical implementation, vs. theoretical decoding that scales exponentially.
Far from rethinking the complete radio, Zyren claimed Intersil "can get to 33 Mbits/s with a CCK/OFDM baseband processor that would drop right into a Prism 2 radio platform."
Zyren also held that patent issues will be resolved for all the applications of OFDM. Though patents might seem like a "scare" point, he said, "the fact that it's taken TI two and a half years to bring a CCK/PBCC product to market should be an even bigger scare point. And there's still no proof it works."
"The patent issues will be resolved as people see OFDM as being the best solution for these kinds of deployments," concurred Nico Van Waef, a Nokia systems engineer who chairs the fixed-wireless-access working group in the OFDM Forum. "People are more concerned about getting OFDM to market now, and will figure out who gets paid what."
Of the trio of proposals, the general consensus is that the M-ary bicoded keying (MBCK) scheme from Supergold is the least complex to implement, and the least harmful in the 2.45-GHz band.
Though it's a startup facing the might of TI and Intersil, Supergold believes the high-rate action opens up room for new players. "Moving to 22 Mbits/s does create more space in terms of opportunities for others to get their coding schemes on the market," said chief technical officer Tim O'Farrell.
MBCK encoding has been criticized for its high peak-to-average ratios, a result of using 16-level quadrature amplitude modulation (QAM). But O'Farrell said that "our analysis shows there's no need to back off any substantial amount more with our power amplifiers than anyone else's, so we don't see that being a problem. In fact, 16-level QAM is even less stressful than OFDM." <<
- Eric - |
