Programmable IC aims to mend wireless LAN muddle

Programmable IC aims to mend wireless LAN muddle
By Patrick Mannion, EE Times
July 10, 2001 (1:06 p.m. EST)
URL: http://www.eetimes.com/story/OEG20010706S0067

MANHASSET, N.Y. — The IEEE 802.11 committee will convene in Portland, Ore., in the next week to grapple with the standards proliferation that threatens to splinter the emerging wireless LAN market. The activity comes as one company tries an end run around the problem with a novel design option.

Systemonic Inc. (Dresden, Germany) is ready to introduce a multiprotocol baseband processor that taps a flexible DSP core to address either the 802.11a or the HiperLAN2 protocol in the 5-GHz band. Future versions will also cater to 802.11b and whatever standards emerge from the work of the IEEE- 802.11g High-Rate Working Group. The technology will complement upcoming RF front-end announcements that will offer both 802.11b and 802.11a radios on a single chip.

The H01 baseband processor addresses a growing demand from global OEMs for a single, flexible platform from which to address wireless communications. Based on the company's proprietary DSP plat form, the baseband chip is supported by an external 64-bit NEC 4332 MIPS processor that handles all media-access control (MAC) layer processing. Systemonic includes the software stacks for both the baseband and the MAC.

In the past, global OEMs have serially designed WLAN solutions for the U.S. standard (802.11a, b or g), then for Europe (HiperLAN2) and then for Japan (HiSwan A, under the Multimedia Mobile Access Communication System, or MMAC), said Rudy Stroh, president and chief executive officer of Systemonic. "Now they can simultaneously design for all regions and upgrade their design in the field as those standards evolve," Stroh said. The result is faster time-to-market and, for the final product, a degree of future-proofing.

Such goals preoccupy designers as WLAN standards continually evolve to address the global market. Analyst Stan Bruederle at Gartner Dataquest estimates that the WLAN sector sucked up between 5 million and 6 million chips last year and that it will consume 10 million to 12 million this year. "In five years, it could be up to 50 million to 60 million [chips] per year, depending on how the home market takes off," he said.

The Systemonic approach is a more immediate alternative to work being carried out by the IEEE's 5-GHz Global Study Group, which is trying to figure out how to get to one global WLAN standard in the 5-GHz band. It also fills many of the needs of an industry advisory group comprising Microsoft, Intel and Compaq. The fractious state of WLAN standards is anathema to that group, whose goal is getting PCs connected.

"Connectivity is an enhancer of these [PC] devices," said Ben Manny, director of residential communications at Intel Corp. "What we're trying to avoid [at 5 GHz] is a repeat of what happened at 2.45 GHz," which is served by both HomeRF and 802.11b.

Wanted: less confusion

Intel had joined the HomeRF camp but recently switched to 802.11b. "We saw HiperLAN2 and 802.11a and made it clear to the industry that we wanted one [standar d], not really caring which," Manny said. But he acknowledged the momentum behind 802.11a, which shares the same MAC as 802.11b.

"What Intel would like to see is less confusion," he said. "And with Bluetooth also at 2.4 GHz, the best thing is for WLANs to move to 5 GHz."

Analyst Allen Nogee at Cahners In-Stat Group agreed with Manny but predicted the 802.11a market will play second fiddle to 802.11b for some time. "It [.11a] has higher capacity and more channels, but we're really worried about its short range — to the extent that it may not work in the home and some enterprise environments," Nogee said. "Most companies we talk to are saving it for the most demanding applications, such as engineering labs and multimedia boardroom presentations."

Nonetheless, there has been a slew of 802.11a announcements of late, building on introductions from Atheros and Radiata late last year. Recent entrants into the 5-GHz 802.11a market include Raytheon, Intersil and Envara (the former hLAN).

But Nog ee issued "a warning to others out there: There's a price war going on in .11a, even before the market has taken off." Atheros has tagged its complete CMOS-based solution at $35. "I suspect that Intersil will be taking a loss at $35 for its solution to compete with Atheros," he said.

Ericsson, meanwhile, will provide the strongest push for HiperLAN2, Nogee believes. The company announced the first products for the HiperLAN2 market earlier this year.

But Nogee said the companies involved in HiperLAN2 are trying to make it all things to all people by cramming too much into the standard. "It's a great standard," he said, "even superior to 802.11a, but it's a case of whether or not they can get it off the ground." While HiperLAN2's supporters wrestle with that, said Nogee, the 802.11a standard is evolving rapidly.

The main advantages of HiperLAN2 rest in its handling of quality-of-service and security features, as well as the European-mandated inclusion of dynamic frequency selection (DFS) and trans mit power control (TPC). But the 802.11 camp, under the respective auspices of the 802.11e, .11i and .11h groups, is addressing those features.

"We're trying to make the DFS and TPC extensions as simple as possible," said Evan Duncan, director of communications architectures at Intel and editor for the IEEE 802.11h Task Group. "We expect it to be an enhancement that will show up in all products, even though it's only needed in Europe," to avoid the need for separate MACs.

Other HiperLAN2 features include a 1394 convergence layer and a 3GPP convergence layer. The former is particularly attractive to the home entertainment market, which is a target of both 802.11a and HiperLAN2 for multimedia streaming.

Smart partnership

Systemonic recently announced a partnership with Sony to address that market, and Gartner Dataquest's Bruederle called that partnership a particularly smart move. Because "802.11a is targeting the home entertainment market," he said, "partnering with the big Japanese consumer manufacturers, including the big set-top-box manufacturers, is essential."

As both standards evolve, the IEEE's 5-GHz Global Study Group continues its efforts to get the two camps to talk to each other. Failing that, the alternative is to come up with another standard altogether.

One option is an internetworking mechanism that would allow peaceful coexistence of both standards in the same physical space, said Bruce Kraemer, senior director of strategic marketing at Intersil Corp. and chairman of the Global Study Group. "Both would recognize which channels the other was using and wouldn't operate on the same channel at the same time," he said. The group is also working on some form of data exchange, "but that's clearly more difficult, with the MACs being so different."

No factions

The second path the group is pursuing is the 5-GHz WLAN Next Generation, where the goal is a single global standard in the 5-GHz space. "This would, in theory, replace both 802.11a and HiperLAN 2 [and the associated MMAC], leaving only one written spec that would be global, without the factions," said Kraemer.

The group is wrestling with the dilemma of whether to break completely free with such a spec or to maintain some level of backward compatibility. It will debate the issue at this coming week's IEEE 802 meeting in Portland.

Also slated to meet in Portland is the high-rate 802.11g group, which is mired in procedural debates arising from the last meeting and in technology debates over the viability of Intersil's orthogonal frequency-division multiplexing solution. Texas Instruments Inc.'s alternative, based on packet binary convolutional coding, was nixed, leaving Intersil's the only option left. But whether it will marshall the requisite 75 percent vote remains to be seen.

"If not," said Intersil marketing director Jim Zyren, "TI's solution might well become a de facto standard. But there have yet to be products announced based on their chip set." Gartner Dataquest's Bruederle sees Intersil's solution as potentially forming the de facto standard.

Sidestepping the debate

In the meantime, Systemonic's H01 offering promises to circumvent the debate by offering a completely programmable solution. The chip is based on the company's VLIW, single-instruction, multiple-data DSP architecture with a configurable data path and extensions for such processing needs as fast Fourier transforms and quadrature amplitude modulation and demodulation. Special interface units get the high-speed data in and out of the core.

The chip has a flexible interface to both direct-conversion and superhet RF transceivers on the front end. "What we're offering is similar to what companies like Tensilica are offering in the microprocessor domain, with a completely automated design cycle and with solutions tailored to the application," said Michael Bolle, Systemonic's executive vice president of engineering.

The chip has a power consumption of 700 mW at full throttle, uses a 0.18-micron proce ss and will be available in the fourth quarter for $28 each in production quantities. Samples are available now for $75.

The company is also offering its HiperSonic Reference Design Platform, which includes the baseband processor, PC card, PCI bus MAC implemented in the MIPS processor, 5-GHz RF front end (the manufacturer is to be announced shortly) and requisite memory (flash and DRAM).