A New York University professor thinks he has found a way for mobile operators to keep data thirsty urban users happy, with the very high radio frequencies that most people had written off years ago used to help carriers stave off a feared bandwidth crunch.
Industry researchers are going to a conference in Brooklyn this week to talk about millimetre-wave wireless, an emerging technology that employs frequencies many times higher than those used today to carry cellular data. Equipment makers are also exploring the frequencies and think they could give an extra boost to the 5G technology coming around 2020, and NYU Professor Ted Rappaport (top image, far right) is a leading proponent of the millimetre-wave wireless.
But Rappaport isn't a typical professor: He's started and sold two cellular technology companies and has more than 100 issued or pending patents to his name, according to NYU. Recently he and his students took to the streets of New York to test millimetre-wave cellular networks and found that they worked better than expected. That's good news, because those higher bands have a lot of spectrum that's only lightly used and may someday relieve a shortage of available frequencies.
If the spectrum crunch that carriers fear becomes reality, it will probably hit first in big cities, where there are often thousands of people using cellular networks at the same time in the same area. The base stations that cellphones communicate with have limited capacity, and the more subscribers share it, the less data each one gets. Eventually, while a simple email check might work fine, watching streaming HD video on a tablet could get iffy.
Millimetre-wave wireless could help delay that problem in at least a couple of ways.
One is to make it easier for a carrier to set up small cells for those densely populated areas. Small cells are like the familiar large cells on towers but sit closer together and serve less territory. They can work alongside the regular cells and deliver more service over the same spectrum. Because they're smaller and there are more of them, it's harder to connect these radios to a wired network. With millimeter waves, those links already can be made using point-to-point wireless beams, which gives carriers more options for where to set up small cells at a reasonable cost.
The other mission for millimetre-wave radios will take a bit more R&D. It involves using the additional frequencies in smartphones and small cells, shooting narrow beams of data around city streets even as mobile users walk and drive around. One way millimetre-wave networks might make this work is by bouncing the beams off the sides of buildings and other objects, which are in plentiful supply in urban areas.
The cellular world has largely stayed away from these frequencies until now because it's hard to make them go a long way, especially when one end of the connection is a moving target like a smartphone. The waves won't go through buildings or cars, as most cellular signals can. Focusing the radio's energy into a narrow beam can solve the distance problem, but typically that requires precise aim.
Rappaport's tests in New York suggest millimetre-wave networks could be made to reach farther than we previously believed. Equipment vendors are excited about those findings, but there will be a lot of questions to address at the conference. Researchers are looking at a variety of possible millimetre-wave bands to use, ranging from 28GHz to 72GHz. More tests are needed to determine just how far millimetre-wave signals can travel. Using those high frequencies could also lead to shorter delays on the network, which would make voice calls and videos work better, but how much shorter isn't yet clear.
Why make the effort? Because millimetre waves offer a couple of advantages that current cellular frequencies can't touch.
First, they reduce interference because the beams are so narrow that there's little chance one will run into another. Second, while there are few frequencies left to harness within today's cellular bands, there are huge amounts of largely untapped spectrum in the millimetre-wave bands.
The cellular industry in the US and much of the world has so far limited its attention to frequencies below 6GHz, NYU's Rappaport said. Looking far above that threshold into bands such as 28GHz, 38GHz, 60GHz and 72GHz reveals wider bands and fatter channels, with more spectrum for all.
"It's really virgin spectrum, very lightly used," Rappaport said. There is an unlicensed portion of the 60GHz band, where the WiGig and WirelessHD technologies are used primarily for in-room video connections and device docking. Other users of millimetre-wave bands include point-to-point wireless backhaul and microwave communications systems that are sometimes used where fiber is hard to install, such as across rivers. But some countries looking ahead to cellular scarcity, including China and South Korea, are already starting to explore these bands for mobile use, Rappaport said.
Poor penetration and other issues make these high frequencies too hard to use without a clear line of sight, analyst Craig Mathias of Farpoint Research said. That's good for backhaul but not for the so-called access networks that talk to phones.
"There are enormous trade-offs here," Mathias said. "I personally don't see a very bright future for the access side of things." Mobile networks have a lot of room to grow just by integrating Wi-Fi, he said. "We have plenty of spectrum down in the more reasonable end of the world where signals propagate a little bit better," Mathias said.
But a new type of antenna, plus a growing amount of number-crunching power, is getting a lot of people interested in millimetre waves. Among the companies participating in the conference are Ericsson, Nokia Solutions and Networks (NSN), AT&T and Intel.
Most of the frequencies used in cellular have wavelengths a few inches long, just about the right size for the mobile phones we carry, Rappaport said. It's possible to make antennas that are smaller than the wavelengths themselves, but that's hard and expensive, so most antennas in cellphones are a few inches long.
Millimetre waves, by contrast, are no longer than a fingernail. That makes it possible to build a new kind of antenna, or rather collection of antennas, that can be precisely aimed and rapidly shifted. This so-called phased array can point in different directions based on the charge applied to a given part of the array. Such directional antennas, in both base stations and devices, allow them to keep sending signals to each other as a user travels. In addition, more than one signal can be steered toward the same mobile device, delivering even more data, Rappaport said.
The NYU professor is on the right track, according to analyst Phil Marshall of Tolaga Research, who once worked with Rappaport at the University of Virginia.
"Your antenna design is what's going to be the key," Marshall said. "How effective you can be in doing very rapid beam-forming." Phased-array antenna technology has been available for years, but for mobile use it takes a powerful processor to keep reconfiguring the antennas so they stay aimed at each other, Marshall said.
At the conference, vendors and researchers will compare notes to figure out just how much they can get out of millimetre waves. That's only the start.
"There are some results available but not nearly enough in different kinds of environments," said Lauri Oksanen, vice president of research and technology at NSN.
Oksanen hopes millimetre-wave networks will be able to reach phones at least 100 metres away. That's roughly how far apart carriers will place LTE small cells, he said. If the higher frequencies provide the same range, then operators will be able to add the future technology at the same sites. NSN's results suggest millimetre-wave cell signals can go tens of metres, at least. Rappaport says his team got good results at 200 metres on the streets of New York.
It's also not clear how soon the new wavelengths will start bouncing around urban canyons. There's already a phased-array antenna on the market for smartphones and tablets, made by Silicon Image for use with WirelessHD. Wi-Fi networks with multiple antennas and beam-forming, available for several years now, use similar technology. The development that's already gone into those systems should help to feed development of cellular systems in the high bands, Rappaport said.
Market forces and regulation will also play a role. Carriers are already committed to rolling out LTE networks for the next few years, so they'll probably wait until at least the next big technology jump to 5G, probably around 2020, Rappaport said. Governments will also have to change their rules for the high frequencies, probably a years-long process.
NSN, which is developing millimetre-wave backhaul technology with partners, says carriers are interested in that technology and the market for it will probably take off as soon as they start rolling out high volumes of small cells. But cells won't start talking to phones over the higher frequencies until later. NSN expects 5G to come out for conventional cellular frequencies first and later for millimetre waves.
"For capacity and supporting different use cases, we don't need the millimetre-wave solution yet by 2020," Oksanen said. "That would be a few years later."
The cellular bands below 6GHz won't suddenly get filled up one day, Tolaga's Marshall said. It will just get more expensive to squeeze the next bit of capacity out of them, until at some point it may be cheaper to use millimetre waves. If that happens, the NYU research could prove prescient, he said.
"Never underestimate Ted Rappaport," Marshall said. "He's a pretty smart guy."