Tech Talk iconTech Talk

A white Starry Beam antenna mounted on a pole for a beta test in Boston, Massachusetts

Startup Says Beaming Millimeter Waves Over the Air Will Make It a Star in Ultra-Fast Wireless Broadband

Standing on the flat roof of a data center at an undisclosed location in Boston, a shivering Chet Kanojia gestures toward a sleek white box about the size of a piece of carry-on luggage. This is the proprietary base station that the seasoned startup founder believes will change the way the world receives its Internet and liberate frustrated customers from the iron grip of legacy providers such as Comcast and Time Warner.

The white box mounted on a pole before us is called a Starry Beam. Only about a dozen of these custom base stations exist in the world right now. The team at Kanojia’s newest startup, named Starry, has spent the past 20 months perfecting the base station’s design. Its performance so far on this nondescript rooftop has persuaded Kanojia that the Internet of the future will not be delivered through expensive fiber optic cables laid in the ground, but beamed over the air using high-frequency millimeter waves.

Standing beneath the Starry Beam, Kanojia points past a spate of warehouses lined with leafy trees to an apartment complex about a kilometer away. There, jutting out from the window of an apartment that Starry has rented, is a white spherical device called a Starry Point. Starry Beams broadcast millimeter waves to Starry Points, which convert them to lower frequencies that flood the home so users can stream ultra-high-definition 4K TV shows to their hearts’ content.

This method, Kanojia believes, can offer much faster service to customers for far less money. In January, he shared that vision at Starry’s launch party in New York City. The company has been quiet ever since, but executives now say their first beta, which has been underway since late August, has confirmed their basic premise—that millimeter waves can deliver ultra-fast broadband speeds up to 1 gigabit per second to customers over the air.

Based on these early results, Starry anticipates average user speeds on its yet-to-be-built network will be as fast as any broadband connection available today—somewhere in the range of 200 to 300 megabits per second. For comparison, the average broadband network in the United States offers average download speeds of just 55 megabits per second.

Right now, Starry’s beta is only measuring the performance of this original Starry Beam that serves a handful of users. In the first quarter of 2017, the company will launch an open beta and build its test network out to a half dozen sites capable of serving several hundred users. Starry has also received permission from the U.S. Federal Communications Commission to run tests in 14 other cities including New York, Dallas, Seattle, San Francisco, and Chicago.

Because Starry CTO Joe Lipowski says the start-up doesn’t plan to publish the results of the beta, it’s hard for anyone to independently evaluate the company’s claims. Starry has not released any press releases about its progress, and Kanojia has also kept the details of his fundraising under wraps. “The less people know about our performance, the better it is for us,” he says.

That attitude has left outsiders wondering what to think of the company’s prospects in such a highly competitive market. “On the surface, the technology sounds like it's sufficient to do what they need it to do,” says Teresa Mastrangelo, a longtime wireless analyst with Broadbandtrends LLC. “We haven’t really seen anything at a big scale. I'll be curious to see how it goes when we're looking at tens of thousands of subscribers.”

If the company can successfully scale, Starry could rewrite the story of what it means to provide high-speed Internet service to homes and businesses. Millimeter waves are high-frequency radio waves that occupy a section of the electromagnetic spectrum that has never been used for consumer technologies. While WiFi, Bluetooth, and cellular carriers have operated on frequencies below 6 gigahertz, Starry is currently testing its technology at 38.2 GHz and 38.6 GHz (where waves are much shorter in length), with future plans to broadcast at 37 GHz and 40 GHz.

Millimeter waves offer several advantages over those delivering cellular data wirelessly on 4G LTE networks and even those carrrying broadband Internet service that is piped to homes through fiber. First, there is a lot more open bandwidth in the millimeter-wave range than there is at lower frequencies crowded with signals from smartphones, microwaves, and WiFi devices. And Starry thinks sending the Internet over the air to consumers will be much cheaper than digging up the ground to lay cables.

In fact, Kanojia estimates that Starry can build out a wireless network that costs only $25 for every home it serves in areas with a population density of at least 1,500 homes per square mile. Installing fiber networks typically costs $2,500 per home. Kanojia thinks the company can make money with market penetration as low as 3 to 5 percent, whereas fiber deployments sometimes require up to 65 percent penetration to be profitable.

One factor that will likely work in Starry’s favor is the range and agility of its Starry Beams. Kanojia says these base stations can deliver superfast Internet service to any customer within 1.5 kilometers who also falls within “near-line-of-sight” of a Starry Beam. That’s an important finding because millimeter waves are often presumed to perform best at shorter distances when there is a clear path between a base station and the end user—and such a direct route can be difficult to find in cities. Millimeter waves can’t easily penetrate windows or buildings, or maneuver around objects like traditional cellular signals can. They are also prone to degrade over longer distances when passing through foliage or rain.

To work around that, Starry equipped each Starry Beam with four active phased arrays, which are rows of tiny antenna elements that cooperate to point and amplify signals in precise directions. With these arrays, a base station can transmit signals more rapidly and with more precision than traditional antennas. In practical terms, this means the Starry network can serve Starry Point receivers mounted on the sides of a building from the same base station that serves those in front by bouncing signals off of buildings and other reflective surfaces. “Our measurements have shown that there’s tremendous reflections,” Lipowski says. Even at what they call “extreme non-line-of-sight” conditions, they’ve delivered data rates of 200 Mb/s to beta users.

Based on these results, Kanojia thinks Starry can provide broadband service with a deployment model similar to existing LTE networks: renting space on existing rooftop cell towers through companies such as the American Tower Corporation. To cover all of Boston, which measures about 230 square kilometers, Kanojia figures the company will need to install three or four Starry Beams at 20 to 30 sites. Each box will support about 1,000 users and boast throughput of 5 Gb/s, for a total of 15 to 20 Gb/s per site. They expect this rate will improve to 45 to 50 Gb/s per site in 2017, once the company upgrades its equipment to meet a new wireless standard known as 802.11ax.

Though Starry says it has cleared some of the biggest technical hurdles that millimeter waves pose for delivering high-speed Internet over the air, it must still find the right pricing model to bring the service to market. “There's no doubt that one could make a system work at 1.5 kilometer range at 37 GHz. In fact, that's a pretty modest range,” says John Naylon, CTO at Cambridge Broadband Networks Limited which operates several millimeter wave networks throughout the U.S. “The issues are going to be economic.”

Mastrangelo, the analyst, says that based on competitors’ rates, Starry would need to price its broadband plan below $100 a month, and ideally between $65 and $85 a month. Unfortunately, Starry’s heavy reliance on custom-built hardware means that its base stations are much more expensive than off-the-shelf models.

Meanwhile, plenty of other wireless providers are rushing to develop their own gigabit solutions. Though Google has more or less abandoned its costly fiber deployments, it recently purchased a company called Webpass that provides wireless broadband to entire buildings by installing rooftop antennas. (Starry offers a similar option for landlords who want to hook up their properties.) Verizon and AT&T have both said that they will launch trials for delivering over-the-air broadband in 2017. Mastrangelo warns that if Starry doesn’t act quickly, the start-up could fall behind.

“If they had been able to come out with a service when they first unveiled it in January, they would have definitely had a huge head start and probably positioned themselves to be an acquisition for somebody,” she says. “But their timing is not fantastic at this stage.”

Jonathan Wells, president of the wireless consulting firm AJIS LLC, says even if Starry can scale and solve the complications of serving hundreds of users at once through phased arrays without causing interference, competition could quickly undercut their plans.

“I think Starry may well be the first there with the technology and if they are successful, they'll get snapped up by Verizon or AT&T,” Wells says. “But I think offering a service that is competitive with Verizon and AT&T is incredibly hard.”

Kanojia says Starry will ultimately compete with its gigabit rivals by providing exceptional customer service, rather than focusing only on high speeds. The company expects to double in size from roughly 100 employees to 200 before the end of next year; among them will be its first batch of customer representatives. But while the Starry team has already proven it can deliver speed, they may find that providing top-notch customer response is more of an art than a science.

The Canadian startup Maluuba has developed deep-learning datasets to train AI on language comprehension and dialogue

Deep Learning Startup Maluuba's AI Wants to Talk to You

Apple’s personal assistant Siri is more of a glorified voice recognition feature of your iPhone than a deep conversation partner. A personal assistant that could truly understand human conversations and written texts might actually represent an artificial intelligence capable of matching or exceeding human intelligence. The Canadian startup Maluuba hopes to help the tech industry achieve such a breakthrough by training AI to become better at understanding languages. The key, according Maluuba’s leaders, is building a better way to train AIs.

Read More
Raytheon Phaser looks like a tan trailer with a flat dish on an arm at top and a rectangle at a 45-degree angle.

Raytheon Sets Phasers to Drone Destruction with Directed Energy Weapon Test

There are all kinds of creative ways of dealing with rogue drones: Radio jamming. Other drones with nets. Trained eagles. None of these are really designed to handle military drones, however, and large, fast-moving UAVs are still a potential threat, especially if more than one is coming at you at once. It's no surprise that the U.S. Army has been developing solutions for this potential threat— we're not sure what they're working on now, but as of late 2013, Raytheon was successfully testing a long range, high power directed microwave weapon capable of taking out swarms of drones in milliseconds.

Read More
Tactical Haptics controller

Tactical Haptics Puts Real Force Feedback in a Handheld VR Controller

In virtual reality games (as in most games), your hands are what you use to interact with your environment, either directly or mediated through some virtual object (like a gun or a sword). But the experience of doing this is almost completely a one-way street: Maybe the controller that you're holding is fancy enough to vibrate a little bit, but that's about the best that you can hope for in terms of the interface physically stimulating you. For a more immersive experience, you want to engage as many of your senses as possible, not just sight and sound.

Besides incorporating motion into VR, adding convincing haptic feedback is the next logical step. It's a difficult step to take, though, because there's no obvious way to exert force on a handheld controller so that it feels like it's responding to the game while it's in your hand. But Tactical Haptics thinks it has the problem licked. It has spent the past few years developing a clever controller that can buck and twist while you're holding it.

Read More

Programmable Millimeter-Wave Mirror Makes VR Wireless

One of the most critical components of a convincingly immersive virtual reality experience is the connection between the headset and the computer. Streaming high-resolution multiview video in real time demands a connection that can reliably handle sustained data rates of more than 6 gigabits per second, and as the resolution and frame rate of VR increases, bandwidth requirements are going to increase as well.

Speeds exceeding 6 Gb/s are easily achievable with a hard-wired connection, which explains why almost all VR systems have a big fat cable tethering the headset to a computer. This isn't an issue for stationary VR, but there are more and more options being introduced that make moving around (at least a little bit) an integral part of VR’s immersive experience. So, being chained to your PC is annoying at best and completely illusion-breaking at worst.

MIT has been working on a way to cut the cord, which sounds like the obvious solution. Current Wi-Fi standards can't handle the amount of bandwidth that VR needs, so MIT researchers have gone another route. They’ve created a system based on millimeter-wave signals (also the foundation of 5G), and directional phased-array "mirrors" that can bounce signals around a room in order to sidestep the line-of-sight issues common to millimeter-wave communications.

Read More
A person wears a white-and-black Sony PlayStation VR headset to play a video game during a launch event in Tokyo, Japan.

The Fuzzy Future of Virtual Reality and Augmented Reality

Virtual reality is having a moment, but the technology is still far from mainstream. Following the release of the HTC Vive and Oculus Rift in early 2016, headset sales quickly flattened once early adopters had purchased their gear. It seems the average customer isn’t as eager to pay US $600 for bulky goggles simply to peruse the rather limited catalogue of available VR content.

If more consumers can’t be persuaded of the value that VR could bring to their lives, an industry projected to breach a trillion dollars in sales by 2035 might flop. Faced with that possibility, NYC Media Lab welcomed virtual reality and augmented reality experts to Viacom last week for a daylong discussion about the future of these two budding technologies.

There was, not surprisingly, plenty of excitement about VR's potential among the crowd of creators, designers, entrepreneurs, and investors. There was also a healthy dose of realism about the state of the industry and the drawbacks of existing VR gear. “For VR to really work and succeed, it has to be so good that you want to put an ugly plastic thing on your face,” said David Liu, creative director for virtual reality at Viacom.

Read More
China and the United States Tied for Number of Top 500 Supercomputers

China and United States Tied for Number of Top Supercomputers

In the latest TOP500 supercomputer ranking, published today, China’s supercomputers are still at the top of the pile—but the United States has caught up in number. Both nations now claim 171 systems in the ranking. And they are roughly equal in terms of raw computing power.

Back in November 2015, China had 108 and the United States 200, or 40 percent—its lowest fraction since 1993, when the list was created. Since then, China has continued to rise, edging past the United States for the first time with 167 systems compared to 165 in the United States in June.

Both nations added new systems to tie in terms of the number supercomputers that rank. They are followed by Germany, Japan, France, and the United Kingdom. China holds 33.3 percent of the total in aggregate Linpack performance and the United States leads slightly with 33.9 percent.

Most of the top 10 supercomputers remained unchanged, with China’s Sunway TaihuLight still clocking in first at 93 petaflops and Tianhe-2 still second at 34 petaflops. Two new supercomputers joined the top 10: the Cori supercomputer at Berkeley Lab’s National Energy Research Scientific Computing Center—skating into the number 5 slot with 14 petaflops—and the Oakforest-PACS at Japan’s Joint Center for Advanced High Performance Computing—taking the number six slot with 13.6 petaflops. Others systems fell to make room, except for Piz Daint at the Swiss National Supercomputing Centre, which maintained the number eight position thanks to newly installed GPUs.

Since last November, the total performance of all 500 computers on the list is 60 percent higher—672 petaflops.

The top 10 supercomputers from the November 2016 list.
Name Country Teraflops Power (kW)
Sunway TaihuLight China 93,015 15,371
Tianhe-2 China 33,863 17,808
Titan United States 17,590 8,209
Sequoia United States 17,173 7,890
Cori United States 14,015 3,939
Oakforest-PACS Japan 13,555 2,719
K Computer Japan 10,510 12,660
Piz Daint Switzerland 9,779 1,312
Mira United States 8,587 3,945
Trinity United States 8,101 4,233

A bright red illustration of a computer chip with the outline of a hacker wearing a hat inscribed on the front.

Wanted: Smart Public Policy for Internet of Things Security

Without us even knowing it, the connected devices in our homes and businesses can carry out nefarious tasks. Increasingly, the Internet of Things has become a weapon in hackers’ schemes. This is possible in large part due to manufactures’ failure to program basic security measures into these devices.

Now, experts in the U.S. are asking regulators to step in. Calls for public policy to improve device security have reached a fever pitch following a series of high-profile denial-of-service attacks leveraged in part by unsuspecting DVRs, routers, and webcams. In October, hackers flooded the Internet service company Dyn with traffic by assembling millions of IoT devices into a virtual botnet using a malicious program called Mirai.

Read More
NYU students Yunchou Xing and George MacCartney, pictured outside of a van in a rural Virginia field on a clear summer day, adjust the horn antenna of their receiver to find the strongest signal during a millimeter wave measurement campaign in August.

Millimeter Waves Travel More Than 10 Kilometers in Rural Virginia 5G Experiment

A key 5G technology got an important test over the summer in an unlikely place. In August, a group of students from New York University packed up a van full of radio equipment and drove for ten hours to the rural town of Riner in southwest Virginia. Once there, they erected a transmitter on the front porch of the mountain home of their professor, Ted Rappaport, and pointed it out over patches of forest toward a blue-green horizon.

Then, the students spent two long days driving their van up and down local roads to find 36 suitable test locations in the surrounding hills. An ideal pull-off would have ample parking space on a public lot, something not always easily available on rural backroads. At each location, they set up their receiver and searched the mountain air for millimeter waves emanating from the equipment stacked on the front porch.   

To their delight, the group found that the waves could travel more than 10 kilometers in this rural setting, even when a hill or knot of trees was blocking their most direct route to the receiver. The team detected millimeter waves at distances up to 10.8 kilometers at 14 spots that were within line of sight of the transmitter, and recorded them up to 10.6 kilometers away at 17 places where their receiver was shielded behind a hill or leafy grove. They achieved all this while broadcasting at 73 Gigahertz (GHz) with minimal power—less than 1 watt.

"I was surprised we exceeded 10 kilometers with a few tens of milliwatts,” Rappaport says. “I expected we'd be able to go a few kilometers in non-line-of-sight but we were able to go beyond ten."

The 73 GHz frequency band is much higher than the sub-6 GHz frequencies that have traditionally been used for cellular signals. In June, the Federal Communications Commission opened 11 GHz of spectrum in the millimeter wave range (which spans 30 to 300 GHz) to carriers developing 5G technologies that will provide more bandwidth for more customers.

Rappaport says their results show that millimeter waves could potentially be used in rural macrocells, or for large cellular base stations. Until now, millimeter waves have delivered broadband Internet through fixed wireless, in which information travels between two stationary points, but they have never been used for cellular.

Robert Heath, a wireless expert at the University of Texas at Austin, says the NYU group’s work adds another dimension to 5G development. “I think it's valuable in the sense that a lot of people in 5G are not thinking about the extended ranges in rural areas, they're thinking that range is, incorrectly, limited at high carrier frequencies,” Heath says.

In the past, Rappaport’s group has shown that a receiver positioned at street level can reliably pick up millimeter waves broadcast at 28 GHz and 73 GHz at a distance of up to 200 meters in New York City using less than 1 watt of transmitter power—even if the path to the transmitter is blocked by a towering row of buildings.

Before those results, many had thought it wasn’t possible to use millimeter waves for cellular networks in cities or in rural regions because the waves were too easily absorbed by molecules in the air and couldn’t penetrate windows or buildings. But Rappaport’s work showed that the tendency of these signals to reflect off of urban surfaces including streets and building facades was reliable enough to provide consistent network coverage at street level—outside, at least.

Whether or not their newest study will mean the same for millimeter waves in rural areas remains to be seen. Rappaport says the NYU team is one of the first to explore this potential for rural cellular, and he feels strongly that it could soon be incorporated into commercial systems for a variety of purposes including wide-band backhaul and as a replacement for fiber.

"The community has always been mistaken, thinking that millimeter waves don't go as far in clear weather and free space—they travel just as far as today’s lower frequencies if antennas have the same physical size,” Rappaport says. "I think it's definitely viable for mobile.”

Others aren’t convinced. Gabriel Rebeiz, a professor of electrical and computer engineering who leads wireless research at the University of California, San Diego, points out that the NYU group ran their tests on two clear days. Rain can degrade 73-GHz signals at a rate of 20 decibels per kilometer, which is equivalent to reducing signal strength 100-fold for every kilometer traveled.

“Rain at 73 GHz has significant, significant, unbelievable attenuation properties,” he says. “At these distances, the second it starts raining—I mean, misting, if it just mists—you lose your signal.”

Rebeiz says signals would hold up better at 28 GHz, only degrading 6 to 10-fold over a range of 10 kilometers. Millimeter waves will ultimately be more useful in cities, he says,  but he doubts they will ever make sense for rural cellular networks: “It’s not going to happen. Period.”

George R. MacCartney Jr., a fourth-year Ph.D student in wireless engineering at NYU, thinks millimeter waves could perhaps be used to serve rural cellular networks in five or 10 years, once the technology has matured. One challenge is that future antennas must aim a signal with some precision to make sure it arrives at the user. That’s because millimeter waves reflect off of objects, and can take multiple paths from transmitter to receiver. But as for millimeter waves making their rural cellular debut in the next few years—“I'd say I'm a little skeptical just because you'd have to have a lot of small antenna elements and you'd have to do a lot of beamforming and beam steering,” he says.

By collecting rural measurements for millimeter waves, the NYU experiment was designed to evaluate a propagation model that the standards group called the 3rd Generation Partnership Project (3GPP) has put forth for simulating millimeter waves in rural areas. That model, known as 3GPP TR 38.900 Release 14, tries to figure out the strength of a millimeter wave signal once it’s emitted from a rural base station according to factors such as height of the cell tower, height of the average user, height of any buildings in the area, street width, and the frequency used to broadcast it.

The NYU group suggests that because this model was “hastily adopted” from an earlier one used for lower frequencies, it’s ill-suited to accurately predict how higher frequencies behave. Therefore, according Rappaport’s team, the model will likely predict greater losses at longer distances than actually occur. Rappaport prefers what’s called a close-in (CI) free-space reference distance model, which better fits his measurements. A representative of 3GPP was not available for comment.

In October, Rappaport presented the group's work at the Association of Computing Machinery’s MobiCom conference and their latest study will be published in the proceedings. In the meantime, it is posted to arXiv.

Planetary Resources’ President & CEO Chris Lewicki and Luxembourg’s Deputy Prime Minister Etienne Schneider celebrate the partnership.

Luxembourg Invests €25 million in Asteroid Mining

Luxembourg has agreed to invest €25 million in asteroid mining company Planetary Resources.

“It’s really big news,” says angel investor Chad Anderson, the managing director of Space Angels Network—one of Planetary Resource’s early investors.

Several companies, among them Planetary Resources and Deep Space Industries, plan on mining space for its riches. Planetary Resources aims to launch the first commercial asteroid prospecting mission by 2020. Near-Earth Asteroids are an untapped reserve of rocket fuel, materials, and minerals, the company explains on its website—although there is some disagreement over whether suitable asteroids are actually available.

Luxembourg was the first European Union country to set up a legal framework for space mining, following the United States Commercial Space Launch Competiveness Act in 2015. In June, the Luxembourg government announced a €200 million fund for enticing asteroid mining companies, as Fortune reports.

In a press release yesterday, Planetary Resources announced that a deal with the Government of the Grand Duchy of Luxembourg and Société Nationale de Crédit et d’Investissement--a banking institution--had been finalized.

€12 million will be a direct capital investment and €13 million will come in the form of grants. Planetary Resources will set up a Luxembourg office, SNCI will take a public equity position in Planetary Resources, and an advisory board member of the initiative will join the company’s Board of Directors, according to the release.

“Asteroid mining is an expensive endeavor,” Anderson says. “Getting this funding is a real benefit to their efforts.”

Planetary Resources and Paul Zenners, Luxembourg’s Ministère de l'Économie, did not respond to requests for further comment.

Amara Graps, a planetary scientist, asteroid mining advocate, and independent consultant for the Luxembourg Ministry of Economy who lives in Riga, Latvia, says “They can put more attention into the asteroid mining business” instead of getting “detoured” by monetary constraints.

Graps says the money is a runway for Planetary Resources to focus on characterizing and identifying proper asteroids as well as refining sensor and propulsion technologies, for example.

She did wonder if this announcement could potentially influence the decision of venture capitalists to invest because of a perceived risk of government involvement.

“Quite the opposite,” says Anderson.

He says the government is only an equity holder with the same stakes as other investors and it would not dictate how the business operates. Because the investment increases credibility, “I think this will encourage other investors to come on.”

“Luxembourg has stepped up,” he says.


Tech Talk

IEEE Spectrum’s general technology blog, featuring news, analysis, and opinions about engineering, consumer electronics, and technology and society, from the editorial staff and freelance contributors.

Newsletter Sign Up

Sign up for the Tech Alert newsletter and receive ground-breaking technology and science news from IEEE Spectrum every Thursday.

Load More