San Jose, Calif. With a freshly updated 802.16a standard in hand, vendors from the WiMAX Forum laid out an ambitious road map last week for delivering merchant chip sets for the first interoperable broadband wireless systems by the end of the year. But the forum's vision of a booming market for high-bandwidth, low-cost systems faces huge design, testing, regulatory and market hurdles.
To overcome them, the group of 68 mostly small OEMs will have to transform this relatively small, fragmented and proprietary sector into one that could compete against digital subscriber line, cable and third-generation cellular in a spate of applications.
At least four digital and four RF chip vendors are racing to carve 802.16a revision D drafted earlier this month at a meeting in Vancouver into silicon by this fall. So far, chip makers are divided on how they will partition their digital chips and what RF interface they will use.
Meanwhile, the forum is gearing up to tackle the job of lobbying for greater harmony in international spectrum regulations governing the 2- to 11-GHz bands in which the products will operate. And it is coming to grips with significant logistics issues regarding how it will test systems intended to deliver up to 74 Mbits/second at distances of up to 50 kilometers.
The most important stumbling block, according to some service providers, may be the latest standard's lack of key mobility and roaming features, which they say they want now. Those features won't be available until a .16e standard is ready, however, perhaps in 2006. That reality is pushing some large and small customers to quietly begin work on a new generation of proprietary systems, even as the forum is trying to recruit a critical mass of carriers to become members.
With the intricate web of issues ahead, many chip and system vendors and carriers are standing on the sidelines to see how 2004 unfolds.
Others are charging ahead toward wireless connectivity, lead by Intel Corp., which promised last July to build a 802.16a chip set by this fall (see www.eetimes.com/story/OEG20030714S0037).
"Wireless is the original multidimensional problem . . . but we don't see any issues that can't be solved by enough smart engineers over time," Sean Maloney, general manager of Intel's communications group, said in a keynote address here last week.
Others were less upbeat. "They will be in a teething stage for the next year or two," said Alie Saad, program director for broadband wireless at market watcher RHK Inc. (South San Francisco, Calif.).
"It will take five years for all these grandiose dreams to be realized," said a more skeptical senior product planner for one large OEM who asked not to be named.
Backers hope WiMAX could be used as a replacement for T1 lines for business, become a backhaul for 802.11 hotspots, provide Internet access for consumers and get integrated along with .11 into notebook computers some day.
One immediate puzzle is how to pair upcoming .16a chip sets from at least four digital baseband and four RF companies. OEMs said some baseband chips will use direct conversion and others intermediate-frequency interfaces to RF. And while Intel and Fujitsu Microelectronics said they will design integrated basebands, others are said to be preparing separate physical-layer and media-access control chips.
"It's a big train wreck, and much more complicated than 802.11a design, especially in terms of dynamic range, spectral range and phase noise in RF," said Colin Howlett, a senior RF engineer at VCom Inc. (Victoria, British Columbia). VCom is an RF subsystem maker that's planning its first WiMAX system.
To negotiate the RF divide, startup Redline Communications (Markham, Ontario) supports both direct-conversion and IF links in the ASICs and FPGAs in its latest system, which implements revision A of .16a. Redline is one of as many as 14 small OEMs in this arena currently evaluating the various WiMAX chip sets. Top-tier OEMs including Alcatel, Harris and Motorola are also conducting WiMAX evaluations.
The Forum plans to finish a .16a test suite for systems by October and hold its first plugfests in November or December. OEMs will then have to undergo final conformance testing, a process that initially could take about a month, at a lab yet to be set up.
"It's a tight schedule, but we think we can hit it," said Paul Senior, vice president of product management and marketing at Airspan Networks (Uxbridge, England). Airspan, which makes fixed wireless systems, plans to be among the first to ship systems using Intel Corp.'s upcoming Rosedale chip set.
"There are a lot of logistics concerns about bringing all this equipment in, setting it up, having the radio licenses you need and testing the stuff in front of a live audience," said Gordon Antonello, senior technical adviser for Wi-LAN Inc. (Calgary, Alberta) and chairman of the WiMAX technical work group. "We are planning to have plug fests every quarter starting late this year so we can make changes in products, tests and standards as we need."
Initially, testing will be conducted indoors only, using wired connections instead of RF. All will be based on .16a revision D, which defines a subchannelization scheme allowing chip makers to use smaller, cheaper power amplifiers. It also provides key hooks for using antenna diversity techniques. In addition, rev D opens the door to chips that can support systems for initial outdoor deployments in early 2005 and indoor deployments in late 2005.
Meanwhile, other forum members aim to lobby governments and standards bodies to harmonize international spectrum rules governing wireless broadband, especially in the 2.5-, 3.5- and 5.8-GHz bands.
"There are many, many issues that need to be resolved so we can make one product that can be sold around the world rather than 57 variants of it for each market," said Senior of Airspan, who chairs the forum's regulatory group.
Not only do spectrum allocations vary from country to country, but requirements for dynamic frequency selection and transmit power control are also inconsistent, especially in the 5-GHz range, he said.
WiMAX backers are also working to repair some disconnects with service providers that want low-cost systems that support portable devices and roaming. The forum plans to support mobility and roaming in 2006-class products based on a .16e standard, which is in the works.
"We're working with various participants to see what we can do to accelerate this," said Sheldon Fisher, vice president of broadband wireless for Sprint, though he would not detail plans. He said that "2006 or 2007 is too late. By that time, 3G will have a strong foothold."
For its part, Nextel is conducting trials for a broadband wireless offering using an undisclosed technology over 2.5-GHz spectrum it purchased last year from Worldcom. The system could be used to deliver data as well as local voice and cablelike video services, said Barry West, Nextel's chief technology officer in a keynote address here.
"I am willing to go with a proprietary technology, but I need to see a path to a standard," said West. He said Nextel could spend up to $2 billion building the network.
Navini Networks Inc. (Richardson, Texas) is developing proprietary broadband wireless technology with portability features. The company has tested the technology with all three carriers in South Korea as well as with Sprint. It plans to migrate its approach to .16e, though it will not support .16a, said Sai Subramanian, vice president of product management.
As for price, West said Nextel needs to deliver a Gbyte of data for $20 or less, a capability it will sell for $30 or more.
Indeed, Sprint's market research shows customers will spend no more than $50 a month for multimegabit wireless access, said Fisher.
The forum is listening closely to carriers and hopes to expand its carrier membership from eight generally small service providers to 24 small and large carriers by the end of the year.
Despite the problems, some carriers are still bullish. "WiMAX is probably the best opportunity we've had in a long time" to get wireless broadband off the ground, said Fisher of Sprint.