Intel announces death of copper
New WLAN standards to deliver wire robustness, scalability
IDF Intel CTO Pat Gelsinger today pledged to enhance WLAN standards and the technologies they encompass to such an extent that copper cabling can be eliminated as a network medium.
"Ultimate vision that Intel has is that we want to make wireless the access technology," he told Intel Developer Forum attendees. "Simply put, [that means] no more copper."
Something of a tall order, you might think, but Intel reckons it can achieve it by steering "future standards", such as the much-mooted 802.11n, to deliver the bandwidth and connection quality we've come to expect from wired connections.
Indeed, Gelsinger pledged that Intel would integrate these developments into the IEEE standards process rather than go it alone as some Wi-Fi vendors have chosen to do.
Bringing wireless up to wired quality centres on a number of techniques, such as using multiple antennae to boost bandwidth and minimising the 'lowest common denominator' factor that drags WLAN performance down to the level of the weakest link.
The multiple antenna system utilises a technique called Multiple Input Multiple Output (MIMO), which spreads traffic across multiple standard-speed wireless channels, boosting overall performance. Improved signal processing yields the superior range. Spatial multiplexing schemes put the data back together again. MIMO uses Orthogonal Frequency Division Multiplexing (OFDM), the basis of 802.11g.
The same technique is already being applied by WLAN chip vendor Airgo Networks. Airgo launched its AGN100 MIMO-friendly chipset last month. One baseband/MAC chip can connect to multiple radio chips and thus multiple antennae.
Interestingly, Airgo's technology sprang from development work conducted by some of its founders at Stanford University. Intel's MIMO also comes from a Stanford, this time via one Dr Paul Raj, a Stanford professor and now an Intel consultant. His company, IOSpan, sold its MIMO IP to Intel.
According to Gelsinger, bandwidth increases proportionally with the number of antennae added to the system. MIMO also makes the system better able to deal with signals that have reflected off objects. Each of these echoes arrive at the receiver at a different time, but MIMO's spatial multiplexing technology can make use of them. This effectively boosts signal strength and so reduces the dependence on getting a good line of sight and improves reliability.
"We're going to drive MIMO into every platform that we build," Gelsinger said. Laptops of the future, he said, will multiple antennae built in, as will PDAs and cellphones.
Intel wants to make MIMO "ubiquitous", he said, and will achieve this in part by "driving it into the key standards of the future". The company is fabbing silicon supporting MIMO today and will bring it to products in the near future, he added. That timeframe suggests that next year's Calexico 2 802.11b/g and a/b/g chipsets are what he has in mind. Ditto products that deliver 802.16a large-area broadband wireless, which will come to silicon next year.
Gelsinger said further bandwidth enhancements will be delivered through adaptive modulation schemes to replace today's "lowest common denominator" schemes that "underutilise the channel dramatically" by giving up "frequencies with a lot of potential bit-loading" to have a "low bit-error rate."
Adaptive MAC systems with packet aggregation and the ability to give transmitters with stronger signals a better slice of the action than transmitters with weaker signals will further boost
As with MIMO, all these techniques will be "driven into the key standards for the future", Gelsinger said - again referring to 802.16 and 802.11a. He expects them to be "widely deployed" in the coming years. Other approaches play into optimising the as yet unratified 802.11e quality of service standard. The IEEE is expected to give 802.11e its stamp of approval mid 2004.
"By employing these techniques, we can make WLANs so robust, so scalable and so cost-effective that we'll eliminate the need for new copper networks for access points in the future," Gelsinger forecast. "That's our goal for next generation of wireless standards.
"We're going to get to a world where all access is done wirelessly - there are no new deployments of copper," he said. Maybe not copper, but there will be fibre roll-outs, he conceded, so we're not looking at an entirely cable-free future. "If you need more bandwidth than wireless offers, move to fibre," he said.
He also admitted this isn't going to happen over night. ®
Intel Developer Forum full coverage
Sponsored: IBM FlashSystem V9000 product guide