A dog’s purpose can take on new meaning when humans strap a GoPro camera to her head. Such “dog cam” video clips have helped train computer vision software that could someday give rise to robotic canine companions.
If they haven’t done so already, cyber attackers may soon be arming themselves with artificial intelligence and machine learning (ML) strategies and algorithms. Before long, it may not be a fair fight if defenders remain naive to what AI and ML can do on both sides of the battle. So suggests a new report by IEEE and the Canadian tech consulting firm Syntegrity.
The report—stemming from a three-day intensive last October of cybersecurity experts from government, the military, and industry—aggregates the group’s findings into what it calls the six “dimensions” at the intersection of AI, ML, and cybersecurity.
In 1981, Richard Feynman suggested that a quantum computer might be able to simulate the evolution of quantum systems much better than classical computers. Except for several proof-of-principle experiments, no working quantum computer has yet been built.
While researchers have succeeded in creating qubits that survive long enough to take part in computations, entangling qubits so that they can form quantum registers that are large enough for any practical purposes has eluded experimenters up to now. Registers temporarily hold data within a processor during a computation.
Results published last week inPhysical Review X by teams of researchers from Germany and Austria have rekindled optimism in the pursuit of a working quantum computer. They report a quantum register of 20 qubits, that, when entangled, can store more than a million quantum states.
When officials at the Federal Communications Commission (FCC) denied launch authorization for four innovative satellites from startup Swarm Technologies last December, the agency was unequivocal as to the reason. “The applicant proposes to deploy and operate four spacecraft that are smaller than 10 centimeters in one of their three dimensions,” read a letter to Swarm’s CEO and founder Sara Spangelo. “These spacecraft are therefore below the size threshold at which detection by the Space Surveillance Network can be considered routine.”
The FCC was worried about collisions in space, where even the smallest objects traveling at orbital velocities can inflict massive damage on satellites or, in a worst-case scenario, manned spacecraft. It thought Swarm’s SpaceBees satellites, measuring 10 by 10 by 2.5 cm, would be just too small to track.
When Swarm launched them anyway, on an Indian rocket in January, the FCC was furious. It rescinded permission for the company’s next satellite launch, due later this month, and questioned whether Swarm was a suitable company to hold a communications license at all. If the FCC comes down hard on Swarm, the company’s ambitious plans for a constellation of Internet of Things communications satellites could be doomed.
So why did Swarm take such a risk in launching its SpaceBees, or even building them in the first place? Perhaps because the FCC’s position on small satellites has been bizarrely inconsistent.
An investigation by IEEE Spectrum has revealed that the FCC licensed multiple satellites smaller than 10 cm over the past five years, including some as small as 3.5 by 3.5 by 0.2 cm. But the commission has also changed its mind from one application to the next, refusing launch permission for satellites that were virtually identical to ones previously authorized. This uncertainty has led to at least one satellite maker exporting his technology rather than risk being denied a license in the U.S.
The Smaller, the Better
Most satellites are essentially smartphones in space. They have sensors such as cameras and magnetometers to gather data, radio transmitters and receivers for communication, processors to crunch the numbers, and batteries to power everything. And like smartphones, satellites have benefited from decades of technological advances such as those whose progress runs on Moore’s Law’s timeclock.
But miniaturization is even more important for satellites than it is for iPhones. Launching a kilogram payload to low earth orbit (LEO) currently costs at least $3000. Multiply that by the 12,000 satellites that SpaceX is planning for its Starlink constellation of communications satellites, and it is clear that lighter and smaller is almost always better.
When Stanford University astronautics professor Bob Twiggs first came up with the idea of a modular “CubeSat” in 1999, he settled on a small 10-by-10-by-10-cm design. Most CubeSats are multiples of this, but their standard 10-by-10-cm footprint allows them to be launched from the same device, regardless of their length.
By the time Twiggs moved to Kentucky’s Morehead State University in 2009, technological advances had accumulated to the point that he proposed an even smaller building block called “PocketQube,” measuring just 5 by 5 by 5 cm.
Four PocketQube satellites were launched on a Russian rocket in late 2013. One of them, the QubeScout-S1 (aka QubScout-S1), was a 5-by-5-by-10-cm sun sensing satellite built by researchers at the University of Maryland. Because of its American roots, the QubeScout is officially a U.S. spacecraft and thus required launch permission from the FCC.
As part of this process, the researchers supplied the FCC with an Orbital Debris Assessment Report (ODAR), a form designed by NASA to assess the danger posed by a satellite to other spacecraft or people on Earth. The standard ODAR covers the likelihood of a satellite hitting something in its orbit, breaking up, or re-entering the Earth’s atmosphere. It does not include any minimum size requirements.
Small as the QubeScout was, Zac Manchester thought he could go smaller. Much smaller. While a graduate student in Aerospace Engineering at Cornell University, Manchester had developed a wafer-thin spacecraft he calls Sprite. It measured just 3.5 by 3.5 by 0.2 cm, and weighed less than 5 grams. Sprites, he imagined, would be modern-day Sputniks, transmitting their names and a few bytes of data to receivers on the ground.
“The idea was to push the limits, to see how small we could really build a satellite,” says Manchester. “One reason is that they can be extremely cheap. We can now talk about launching a satellite for a couple of hundred bucks.”
In 2011, Manchester launched a Kickstarter campaign to build a CubeSat mothership that would deploy 128 Sprites into low Earth orbit. He quickly raised nearly $75,000 and secured a position on a NASA-sponsored launch due for 2014.
“We thought about tracking and debris concerns from the outset,” says Manchester. “We got vetted by the NASA Orbital Debris Program Office (ODPO), the people who spend their lives dedicated to making sure we don’t make Earth orbit uninhabitable. We also spoke with the Department of Defense and the Joint Space Operations Center who operate tracking radars.” Manchester even ran a statistical Monte Carlo analysis to determine the collision risks.
“At the end of the day, everyone signed off on it and said you guys are OK,” he says. The FCC did too, authorizing Manchester’s KickSat satellite in the spring of 2014 without raising any concerns.
In the summer of 2014, the FCC licensed yet two more tiny satellites: AeroCube-6A and -6B. They are a pair of 5-by-10-by-10-cm satellites designed to measure space radiation impinging on spacecraft. These small satellites, dubbed “picosatellites,” were produced by the Aerospace Corporation, a federally funded R&D center for the U.S. Air Force’s Space and Missile Systems Center. Like the PocketQubes before them, the AeroCubes were launched and are still being tracked successfully. Picosatellites seemed to be flying high.
A Kick in the Sats
But space can be cruel. Although Manchester’s KickSat mothership launched perfectly on a SpaceX Falcon 9, it was released in an extremely low orbit where atmospheric drag would cause it to re-enter and burn up in a matter of weeks. The Sprite deployment then had to be delayed by 16 days to allow for the possibility of a manned Soyuz mission to the International Space Station going wrong. (If the mission aborted, its planners would not want the capsule to return to Earth through the cloud of tiny satellites). While the Soyuz mission went off without a hitch, something went wrong aboard the KickSat in the meantime and it burned up without having released its Sprites.
Luckily, Manchester had enough spare parts to piece together a replacement mothership and a second flock of Sprites, and NASA said that it would support another mission. Manchester submitted a new FCC application in 2015, almost identical to the first.
As before, the tiny Sprites would operate for less than a week before burning up in the atmosphere and, as before, Manchester had gotten the go-ahead from NASA’s space junk experts. “For KickSat-2, releasing 100 Sprites from 325 km altitude has no long-term negative impact to the environment. The ODPO is fine with the plan,” wrote Jer-Chyi Liou, NASA’s Chief Scientist for Orbital Debris. But this time the FCC’s response was very different.
Towards the end of May 2016, Manchester was preparing the KickSat-2 mothership in Ithaca, New York, for its flight to Houston’s Johnson Space Center, where it would be integrated into the launch rocket. The weekend before he was due to fly, Manchester got an email and then a phone call from the FCC, saying that the agency was not going to licence the satellite.
The letter from the head of the FCC’s experimental licensing branch spelled out its thinking: “We are not able to approve your request due to concerns regarding the overall risk and uncertainties associated with your experiment. There are concerns about the trackability and operational impacts of these satellites (Sprites), and of this type of satellite in general.”
“This is a crazy thing to happen two days from delivering a satellite,” says Manchester. “We had already been vetted by the NASA office that actually does this for a living. It wasn’t a problem before and it’s not really a technical problem. It’s more of an unfortunate bureaucratic situation... regulation gone awry.”
While Manchester attempted to reason with the FCC, his NASA contacts tried to pull strings within the federal government, but to no avail. As months ticked into years, Manchester realized that an FCC license might never be forthcoming.
Eventually, NASA adopted KickSat-2 as an official NASA mission. Due to a regulatory quirk, NASA’s own satellites are overseen by the National Telecommunications and Information Administration (NTIA) rather than the FCC. KickSat-2 is now slated for launch late this year or early next. “But it’s taken us three years and lots of lawyers to figure out,” says Manchester.
IEEE Spectrum could find no record of any satellite smaller than 10 cm on any side having been licensed for launch by the FCC since the KickSat-2 fiasco. We gave the FCC multiple opportunities to explain its inconsistent approach to licensing small satellites, but it has yet to reply.
Tom Walkinshaw, founder and CEO of Alba Orbital, a start-up based in Scotland that develops and sells PocketQube satellites, calls the FCC’s erratic picosat licensing decisions “a head-scratcher.” He says, “Is one part of the organization acting one way, and another part another? I don’t think anyone’s too sure right now. Everyone’s looking for clarification.”
The FCC attempted to shed some light on its thinking last month, with a proposed rule-making for streamlining the licensing procedures for small satellites. The main aim of the new rules, welcomed by every expert IEEE Spectrum spoke to, is to reduce the paperwork and time needed to license the launches of small, experimental satellites and CubeSats. But the rules explicitly exclude picosats, requiring satellites to be no less than 10 cm on any side “to ensure that the satellite will be trackable as a space object.”
Walkinshaw points out that not only are all of today’s PocketQube satellites (and Swarm’s SpaceBees) being successfully tracked by existing technology, a new S-Band “Space Fence” due to be deployed by the United States later this year will explicitly monitor objects with dimensions as small as 5 cm.
A bigger problem, thinks Manchester, is that “the FCC has been unwilling to consider licensing anything smaller than a CubeSat through the normal application process, and this new process does not seem to address that broader issue in any way.”
When the Breakthrough Starshot initiative approached Manchester in 2015 offering to develop his Sprites into miniscule interstellar spacecraft for a journey to Alpha Centauri, it wanted to test them in Earth orbit first. “We didn’t even try to go through the FCC,” says Manchester. “We got an export licence from the State Department, which took a couple of weeks, and then I just took the satellites to Germany in my carry-on.”
Last summer, the London-based Breakthrough launched six Sprites on a German satellite from an Indian launch vehicle, far outside the FCC’s jurisdiction. Germany has no minimum size regulations for satellites.
“I want to be responsible around debris concerns,” says Manchester. “But because of the situation with the FCC, it’s now advantageous for a company that wants to [work with small satellites] to relocate and do it as a foreign company. And that’s probably not great for the U.S. long term.”
Update 15 April 2018: After publication, Neil Grace, a spokesperson for the FCC, sent this statement in response to an earlier request for comment: “Size isn’t the only criteria that the FCC considers when granting experimental licenses for small satellites. Due to the number of variables to consider, the Commission takes a wholistic case-by-case approach when granting authorizations for satellites of this size.”
For 5G, researchers need city-size playgrounds in order to properly test and develop their technologies. That’s why the National Science Foundation (NSF) announced today that it will deploy two Platforms for Advanced Wireless Research.
The two PAWR (pronounced “power”) test beds will be in Salt Lake City, Utah, and New York City. For the Salt Lake test bed, the NSF will partner with the University of Utah and Rice University (in Houston, Texas) to oversee the platform. In New York, Rutgers University, Columbia University, and New York University will have that responsibility.
The sheer scale of these test platforms will allow for research options previously unavailable to researchers. “We’ve made limited measurements in the laboratory,” says Sundeep Rangan, the director of NYU Wireless, “but nothing at this scale.” Rangan says the platforms will allow more thorough testing of some of the big promises of 5G, including applications such as VR/AR and autonomous vehicles.
These test beds are about more than testing 5G, though. “What we are thinking is after 5G,” says Thyaga Nandagopal, the deputy division director for the Division of Computing and Communication Foundations at the NSF. In years to come, these test beds will facilitate research into bands of spectrum that have never been used for telecommunications.
Solar cells convert light to electricity. Image sensors also convert light to electricity. If you could do them both at the same time in the same chip, you’d have the makings of a self-powered camera. Engineers at University of Michigan have recently come up with just that—an image sensor that does both things well enough to capture 15 images per second powered only by the daylight falling on it.
With such an energy-harvesting imager integrated with and powering a tiny processor and wireless transceiver, you could “put a small camera, almost invisible, anywhere,” says Euisik Yoon, the professor of electrical engineering and computer science at University of Michigan who led its development. They reported their results this week in IEEE Electron Device Letters.
A diving trip to the Great Barrier Reef may have unlocked a new way to build a GPS-like sensor that works underwater. The device is based on recent scientific understanding of how marine animals sense their geolocation based on the signature polarization patterns of light entering the water.
It’s 2018 and more than 16 million people living in the rural United States still lack adequate access to mobile broadband. But building out that infrastructure is an expensive endeavor. One analysis from 2017 estimates it would take 37,500 new cell phone towers and run upwards of $12.5 billion to bring 4G to rural areas in the United States, including Alaska, Hawaii, and Puerto Rico.
The team at Altaeros, a Massachusetts-based company, thinks they have a better plan. Their SuperTower platform employs a tethered, autonomous aerostat that can lift antennas and receivers to an altitude of 250 meters (820 feet) to deliver mobile broadband to underserved communities. One aerostat can provide coverage for up to 10,000 square kilometers (3,860 square miles), an area that would normally require between 20 and 30 cell phone towers. As a result, the cost of deployment is about 70 percent cheaper than the conventional infrastructure, says Ben Glass, CEO of Altaeros.
Since its invention in 1960 by Charles Townes, the laser has become as ubiquitous in today’s technology as the transistor or the microprocessor. Townes was also the inventor, in the early ‘50s, of the maser, the progenitor of the laser. Instead of light, it amplified radio waves, and its operating principle, stimulated emission, became the working principle of both the maser and laser.
The maser had an interesting property: it could amplify radio signals without adding noise caused by the motion of electrons. Radio astronomers turned to masers to amplify weak radio signals from distant galaxies and quasars, and NASA used them in its Deep Space Network to communicate with the Voyager space probe.
However, unlike the laser, the maser’s applications remained limited, mainly to atomic clocks and for metrology. One reason was that masers require refrigeration with liquid helium to temperatures close to absolute zero, which increased the bulk of these devices substantially. Another reason was that during the 1960s, cryogenically cooled low-noise FET microwave amplifiers and oscillators that were simple and flexible in their design and capable of achieving very low noise temperatures, also became available.
Now, a team of researchers from the Imperial College London, the University College London, the Queen Mary University London, and the University of Saarland in Saarbrücken, Germany have reported in Nature a maser that does not require cooling and might bring back the maser as a low-noise amplifier. Instead of rubidium and other gases, or ruby sapphire crystals, they used diamond to amplify the microwaves.
MicroLED displays are screens built from tiny versions of the same sort of gallium nitride chips you find in LED lights. They promise double or triple the power efficiency of today’s OLED and LCD screens and brightness that is orders of magnitude better. So, it’s no surprise that both a crowd of startups and at least one gargantuan gadget-maker are all racing toward making the first commercial screens.
Most startups have chosen to develop so-called monolithic displays, which build the display as a single chip or two chips bonded together. It’s potentially a shorter route to success, but it’s likely going to be useful only for very small screens such as those needed for augmented reality gear.
Apple, with its deeper pockets and early start, is going for a technological high-dive with a higher degree of difficulty. Its formulation could be used for smart watches and ultimately even larger formats. One startup thinks it can match or maybe beat Apple to the ultimate prize: a microLED smartphone screen.
Such a smartphone would last at least two days without a recharge if today’s usage patterns hold up, says Reza Chaji, CEO of Waterloo, Ont.-based startup VueReal. But at UHD resolution, smartphones are the most demanding application for microLEDs, he says. Success with a smartphone, though, would be a “New York, New York” kind of moment. If you make a microLED smartphone work, you can make a larger screen of any size with the same or even less difficulty. And that’s why all of VueReal’s efforts are focused on the smartphone, says Chaji.
Unlike in monolithic displays, the pixels in the screens VueReal and Apple are chasing are made up of individual LED chips. And that leads to three big manufacturing problems. First, LEDs are fantastically efficient at the size used in lighting, but shrinking them down to micrometer-scale saps their efficiency. LEDs have border areas that leak current that could otherwise go to making light. As the chips get smaller, the ratio of border to light-emitter gets closer to 1:1 and efficiency sinks like a pair of lead swim trunks.
Second, displays are unforgiving of errors. A yield of 99 percent might sound good, but it’s belly-flop-on-national-TV bad in displays. For UHD resolution, it would mean as many as 250,000 dead pixels, which is something you’d definitely notice.
And finally, there’s the question of how you get 20 million individual LEDs into their respective spots on a screen in something less than several weeks’ time.
According to Chaji, VueReal has solved all three problems. But it’s the last that’s perhaps most crucial and the one that offers the most contrast to what observers assume Apple is doing. Though Apple won’t talk about its microLED display manufacturing technology, those who have examined the company’s patents believe it is working on a sort of rapid “pick-and-place” mass transfer technology, where LEDs are delivered from a batch and individually put in place on a substrate.
Chaji won’t reveal how VueReal’s technology, called solid printing, works, but says “it enables us to do a lot of parallel transfer into a display, and it could populate a TV in less than 10 minutes.” He says it is more like photocopying or document printing than a pick-and-place. Those using pick-and-place “might be ahead of us because they started sooner, but in the long term I think our solution will surpass pick and place,” says Chaji.
The company plans to be ready for full production in 2021, when it will be able to supply technology, equipment, and materials to partners.
This post was corrected on 29 March 2018 to clarify the relation between VueReal’s technology and what is thought to be Apple’s.
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.