PALO ALTO, Calif.—Intel Corp.s communications chief tempered expectations of a future “Radio Free Intel” project, warning that it will be some time until a single chip will be able to access a variety of wireless networks and cellular services.
However, an Intel spokesman said an announcement on the companys Wi-Fi-to-cellular roaming strategy is expected in two months time, and roadmaps presented here at a wireless analysts meeting on Friday showed that the company would release a “wireless WAN” chip in 2005.
In addition, the director of Intels communications lab disclosed that the company has implemented a 10GHz frequency synthesizer in CMOS silicon, a key step on the way to building a hybrid controller for both Wi-Fi and cellular networks.
With a 10-GHz synthesizer, Intels processor will handle both the 5GHz bands required by 802.11a and 802.11g Wi-Fi radios as well as the lower frequencies used by the next-generation 3G protocols.
“Were taking the approach that 10GHz covers all the bands of interest, said Krishnamurthy Soumyanath, director of the communications and circuits lab at Intel.
The concept of a “Radio Free Intel” was first introduced in the fall of 2001, when Intels chief technology officer Pat Gelsinger described the concept as his personal research crusade. The concept includes several components, including smart antennas, regulatory reforms, and a CMOS-based reconfigurable radio.
His idea was that a radio could be programmed via software to tune into a variety of wireless networks that use different frequencies, such as 802.11a, 802.11b, and cellular networks like W-CDMA.
However, Sean Maloney, executive vice president and general manager of Intels communications group, said the concept of a reconfigurable radio “superchip” was now considered unlikely. “Im trying to not sound like a hippie, but I think that it is the journey, not the destination,” he said.
Maloney said that the concept of a software-defined radio is not dead, but that the company was leaning more towards “flexible” radios, the implication being that Intel instead will try to cover a relatively limited range of spectrum with one chip.
Mike Chartier, the companys director of spectrum policy, also used the term “flexible”, claiming that the company was working with the International Telecommunications Union to develop local rules that would govern the flexible radio technology by next year.
Executives from Intels wireless and mobile computing business discussed Intels strategy on several fronts here, including lobbying efforts to free up wireless spectrum, improvements in the design of mobile PCs, as well as a report by Intels resident anthropologist on the impact of mobile devices on Asian society.
Beginning with Intels “Unwired” advertising campaign surrounding its Centrino wireless platform, Intel has made wireless communications a key element of its evolving business model. The companys lofty goals have contrasted with its execution, however, as the Intel Communications Group has reported a string of unprofitable quarters.
Furthermore, although Intels Centrino dominates the thin-and-light notebook platform, most early Centrino notebooks shipped with third-party Wi-Fi chip sets. Intel also delayed the wireless daughtercard associated with its Intel 915 or “Grantsdale” chip sets, which analysts interpreted as a lack of available radios.
Rob Crooke, Intels vice president of the desktop platforms group, explained away the chipsets delay by claiming that its year-long development time “didnt allow enough time to test the user experience” and that the chipset would be ramped during the second half of this year.
Hybrid Wi-Fi chips are nothing new, microprocessor analysts have observed, noting that Intel has been slower to market with these products than its competitors. In addition, Taiwan chip vendors with low cost chips are now entering the market with home-grown Wi-Fi chip designs, whose entrance has traditionally been a death knell for profit margins in the electronics industry.
Meanwhile, a number of other Intel wireless efforts have begun, seemingly in isolation. Intels other maturing wireless line lies in handheld wireless networks, developing applications processors that can be used in PDAs and smartphones.
In the past, Gelsinger has described an environment where a laptop user would surf on a Wi-Fi network at work or in a café, and seamlessly shifting to a cellular network while driving down the highway.
However, Intel sidestepped the IEEE standards body in developing a short-range UWB (ultrawideband) standard, and has been the chief promoter of WiMax, a next-generation wireless broadband standard which will slowly emerge in the next few years.
Soumyanath said that Intel is working on sort of a reverse-branching design path, where various functions are integrated over time. For example, Intel and other companies have designed discrete 802.11b chips, then later added other Wi-Fi functions as 802.11a.
Intels 802.11a/g wireless components are now being sampled by OEMs, and will ship in products this fall, according to Jim Johnson, vice president and general manager of wireless networking for Intels Communications Group.
Hybrid chipsets, 10GHz
-band support on the way”>
Questions still remain on when Intel will manage to ship a hybrid chipset that bridges Wi-Fi functionality with cellular functions, such as the next-generation CDMA2000 3G telecommunications standard adopted by U.S. providers and the more common W-CDMA standard that European and Asian carriers have adopted. The Friday briefing suggested that may happen in 2005, where Intels roadmap shows “WWAN” components being offered to handset vendors.
A wireless chip set typically consists of at least two components: the digital controller and the analog radio. Code-name “Hermon,” Intels next-generation embedded controller, will be used to control an external W-CDMA radio and is due to ship next year. Intel has not said whether or not it plans to actually manufacture the 3G radio component itself, spokesman Manny Vara said.
Typically, a radio is a “dumb” component, tuning itself to whatever frequency the controller demands and passively transmitting and receiving the encoded data. Intels research is currently focusing on agile radios that can communicate over a variety of bandwidths and frequencies, said Ben Manny, director of the radio communications lab within Intels corporate technology group.
Future radios, however, will also include cognitive elements to sense and adapt to the available frequencies, such as cellular or Wi-Fi networks. Future reconfigurable radios will actually reconfigure themselves to adapt to various radio schemes, he added.
The hybrid wireless picture may grow clearer in two months time, when Intel will make an announcement regarding roaming between Wi-Fi and cellular networks, spokesman Manny Vara said. He declined to provide more details. The idea is that various wireless networks will coexist, not compete, he and other Intel executives said. From a development standpoint, that mandates a chip that can interact with all of the various networks.
Meanwhile, Intel has begun developing the software infrastructure needed to manage the evolution of hybrid wireless chipsets.
Intels Adaptive Radio Architecture combines two layers: one designed to manage the radios themselves, and an applications layer designed to facilitate functions like handing off a voice-over-IP call to a cellular network. The Adaptive Radio Architecture is being built in to Intels Personal Communicator, a proof-of-concept device that is designed to access streaming multimedia content both over a Wi-Fi LAN as well as a cellular network.
“The challenge is how we can use the best of those networks to deliver the best user experience,” said Bryan Peebler, senior programs manager for Intels Systems Technology Lab.
“This is designed to be able to replace discrete radios and replace them with a small number of programmable radios,” Peebler added.
At the end of the day Friday, executives showed the progress on the implementation of a 10GHz frequency synthesizer in silicon, which could dial down to a variety of lower frequencies through the use of non-integer divisors. This capability could be critical, as W-CDMA covers both the 1,920MHz-to-1,980MHz and 2,110MHz-to-2,170MHz frequency bands.
A collection of devices, in close proximity, all broadcasting at 2,150MHz, for example, would interfere with one another, he said. The oscillator can hop from frequency to frequency, supporting Intels “agile radios.”
Soumyanath also said that Intel had developed a complete 802.11a transceiver in 90-nm CMOS silicon, as well as a 100GHz oscillator in the same process technology. Like the frequency synthesizer, the oscillator and 802.11a component are designed to show that wireless components can be manufactured on 90-nm silicon.
Still, Intel and other chip vendors have suffered growing pains with 90-nm fabrication. The company recently delayed the release of 4GHz Pentium 4s produced in the 90-nm process by at least a quarter.