Wireless inside a wired Swiss cheese
SAN JOSE, Calif. – At a time when carriers are pondering an expensive shift from copper cables to optical fibers, a pioneer of digital subscriber lines is proposing a novel upgrade that someday could deliver terabit data rates.
In a keynote at the G.fast Summit in Paris Tuesday (May 9), John Cioffi will unveil ideas behind what he calls Terabit DSL (TDSL). They include carrying 50-600 GHz wireless signals through the tiny spaces between individual twisted pairs or the cables that bundle a hundred of them.
“We are shooting for a terabit/second over 100 meters, 100 Gbits/s at 300 meters and 10 Gbits/s at 500 meters -- all those are 200 to 1,000 times better than traditional DSLs,” said Cioffi, whose research at Stanford in the 1980s led phone companies to embrace DSL for broadband.
“Those numbers could be off 10-25 percent, but even if they are off by 10x, it’s still a substantial improvement—there’s a lot of opportunity here,” said Cioffi who is now a professor emeritus at Stanford and chairman of Adaptive Spectrum and Signal Alignment Inc. (ASSIA), a developer of DSL software.
The technology could be used to mitigate the need for running fiber to the home as well as to carry traffic from 5G cellular base stations. Cioffi has created computer models he and others believe prove the concept that could take at least two to three years to commercialize, he estimated.
Carriers charge as much as $4,000 to pull fiber to an individual home, a cost that could come down to as little as $200 for a shared fiber/TDSL link, Cioffi said. Others estimate it could cost 400 billion euros to deploy 5G cellular throughout Europe, a price tag that could be reduced to tens of billions using TDSL backhaul, he added.
“The largest cost in 5G is not the wireless part, it’s the wires behind it,” said Cioffi, who posted online slides from his TDSL talk.
TDSL sends 50-600 GHz wireless signals in gaps between copper pairs. (Images: John Cioffi)
The concepts in TDSL had their start two years ago when Cioffi was reviewing 60 GHz MIMO work at a wireless lab at NYU led by Ted Rappaport. “I started thinking whether there was an analog for this work in cables, I talked to Ted and he encouraged me,” Cioffi said.
That talk led to discussions with other academics and transmission specialists at Apple and Huawei. About a dozen colleagues have reviewed Cioffi’s work so far.
"TDSL is a breakthrough in the pursuit of ultra high speed over copper. The combination of millimeter waveguides and vectoring signal processing puts new regimes of physics together with advances in mathematics--the business implications of this technological breakthrough will be tremendous,” said Mung Chiang, a professor of electrical engineering at Princeton.
"Surface wave propagation is an exciting new transmission mode that could be harnessed for massively broadband communication. I wouldn't bet against John Cioffi, he's the pioneer who brought us DSL,” said Ted Rappaport.
Next page: How it works and what’s next
How it works and what’s next
TDSL antennas may look like bow ties or donuts.
Cut sideways a typical bundle of twisted pair lines looks a Swiss cheese. TDSL uses those gaps to send wireless signals. Wires don’t carry signals directly but act as waveguides for them.
Because the gaps are so small, the signals need to use tiny wavelengths. “We used 50 GHz to 500-600 GHz, it’s not all going to work on any one cable,” Cioffi said of models he developed with colleagues at ASSIA.
“Instead of injecting current in the wires, he foresee putting almost microscopic antennas like a donut or bowtie around the wire--not touching it,” Cioffi explained. “Shooting an IR laser at those elements creates a millimeter wave similar to what’s done in coherent optics, but we use a different frequency and an array of signals instead of just one used in optics."
The array of antennas applies concepts well studied in MIMO wireless and vectored-DSL wired systems for “steering messy, multi-dimensional channels,” he said.
The good news is adequate silicon support is here or on the horizon. Existing 120 GHz ADCs and cost reduced versions of today’s 16nm GPUs could handle the electronics, he estimated. Nevertheless, much work is still ahead.
“At these speeds devices to get signals in and out of the Swiss cheese will be expensive to develop...it’s a multi-million project for someone to measure” the TDSL effects with high accuracy, he said. “I’m trying to encourage as many large service providers and big network equipment providers to validate [the work] and add their own contributions,” he added.
“If there’s a motivated entity, this can be done in two to three years, it’s a matter of someone believing they will get a return and making it happen in a way that can be mass reproduced…[but] it’s too early to project the best business models,” said Cioffi who in the 1990s formed Amati, a startup that made ADSL chips and was ultimately acquired by Texas Instruments.
At a time when companies across the industry are collaborating to spend hundreds of millions of dollars on 5G cellular, “I’d love to see that investment and intensity here, the return could be even larger when you consider delivering bandwidth to every cell site, let alone home,” he said.
— Rick Merritt, Silicon Valley Bureau Chief, EE Times