Autonomous cars drive terahertz research

Self-driving cars are zooming towards reality, and one day in the not-so-distant future people may glide down city streets as passengers, occupied with activities other than keeping their eyes on the road. The kids could be playing video games, while mom participates in a work meeting using the 2030s equivalent of Zoom.

Meanwhile, the car will be assessing its environment with visible light and infrared cameras, measuring the distance to other vehicles with lidar, and communicating its intended movements to neighboring cars. Traffic signals will measure oncoming vehicles and speed them up or slow them down to assure everyone gets through the intersection safely with minimal waiting, while nodes on buildings or streetlights collect data about driving patterns and send it to a centralized controller, which in turn sends back information to reroute traffic to reduce congestion or avoid construction sites.

That's an awful lot of data being sent and received, whether from sensors to processors or vehicle to traffic central, and it all has to pass swiftly and accurately to avoid traffic jams and crashes. To achieve that level of communication, new technology is being developed, and it may well rely on the nascent field of terahertz photonics.

In a swiftly changing environment, such as an intersection full of self-driving cars, "you need to go to these very fast communications systems," says Idelfonso Tafur Monroy, a professor in the Terahertz Photonic Systems Group at Eindhoven University of Technology in the Netherlands.
"We are looking into how you can do this with sub-terahertz or terahertz wavelengths."

He cites an estimate from a company that builds high-frequency antennas that the average self-driving car could generate roughly two petabits of data every week, about enough to completely fill the hard drives of 250 laptops. A 4G wireless system would take 200 days to transmit all that data, he says, by which time much more would have been generated.

Higher frequencies, though, mean they can accommodate higher data rates. Terahertz frequencies could easily transmit a terabit of data—one trillion bits—every second, and perhaps much more. That's 50 times the data rate of a 5G connection, which makes the waveband attractive for autonomous vehicles or so-called smart highways, where data is coming fast and furious from multiple moving sources. A similar range of frequencies might also be used in sensing, an important component of autonomous driving, meaning the potential exists for combining two systems into one, simplifying the whole project.

Terahertz waves occupy the part of the electromagnetic spectrum between far infrared light and microwaves, on the cusp between optics and radio, where people stop defining radiation in terms of wavelengths, and instead talk of frequencies. The International Telecommunication Union defines the terahertz band as running from 0.3 to 3 THz, although it is sometimes considered as anything from 0.1 to 30 THz.

It's also the region where the waves might be generated either electronically or photonically, and there's not widespread agreement on which will work best. "The terahertz band is in between the comfort zone of people in electronics and the comfort zone of people in optics," says Josep Jornet, a professor of electrical and computer engineering who directs the Ultrabroadband Nanonetworking Laboratory at Northeastern University in Boston.

"People in electronics are trying to get to the terahertz band by pushing frequencies up," Jornet says. "But then you have the people in optics who say, well, terahertz waves are nothing but weak photons."

Network scenarios using the combination of different technologies based on 5G, light fidelity (LiFi), and terahertz (THz) systems for autonomous driving vehicles. Credit: Idelfonso Tafur Monroy/SPIE

SUBHED

While electronics researchers can create circuits to double, triple, or quadruple the frequency of their signal, photonics engineers can use similar tricks to go in the opposite direction. They can take two telecommunications lasers with slightly different wavelengths and superimpose their beams on a photodetector, producing a "beat note" that is the difference between the two. Or they can train a laser on a photoconductor connected to a metallic antenna to produce a terahertz signal.

At the Center for Converged TeraHertz Communications and Sensing (ComSenTer), a consortium of research universities working on next-generation wireless communications and led by the University of California, Santa Barbara (UCSB), research focuses on systems operating between 100 and 300 GHz. Mark Rodwell, a professor of electrical and computer engineering at UCSB who directs ComSenTer, says it's more accurate to call that the upper portion of the millimeter wave frequency band than the lower end of the terahertz band. And he says the center doesn't get into photonics, "as transistors and integrated circuits work extremely well in this frequency band." He and others have used transistors to reach even higher frequencies, and Northrop Grumman, the aerospace and defense contractor, demonstrated a transistor-based circuit that could reach 1 THz back in 2014. "The use of photonics for these applications, in my technical judgment, is of very limited utility," Rodwell says.

Jornet says the 100 to 300 GHz range is of interest because the technologies that use them are probably closer to being commercialized. Frequencies closer to 1 THz could also prove valuable, he says, and photonics can play an important role in developing them for communications. "The fastest communication systems that we have access to are usually based on optical fiber," he says.

Communications systems have to modulate the signals they send out in order to imprint data on them, and that's something photonics experts are used to. Researchers increasing the frequencies of transistors will have to develop new ways to modulate those signals very quickly. "If you already have technology that can manipulate and handle very high-speed signals, for example, optical
communication systems, and then you can reconvert those to terahertz frequencies, well, you're halfway there," Jornet says.

Some researchers argue there's no point in trying to use frequencies above 300 GHz, because there's too much atmospheric absorption. But Jornet says that, while there certainly are areas between 300 GHz and 1 THz where much of the signal would be absorbed, there are wide bands where there is very little atmospheric loss. "Yes, there is absorption, but not at every frequency. So you need to be smart when picking the frequency," he says.
Both electronic and optical approaches offer their own advantages, says Guillaume Ducournau, a professor in the terahertz photonics research group at the University of Lille in France. "The highest data rate has been achieved with photonic-driven solutions, and the highest transmission distance has been achieved with electronics, but generally with a lower data rate," he says.

Photonics, while allowing for fast modulation of the signals, has trouble reaching high powers at these frequencies. The most advanced sources have output of approximately 1 milliwatt at 300 GHz, which limits the effective distance of the transmission. Depending on how the system is designed, that may provide an ability to send data distances of only a few meters or tens of meters, enough for car-to-car communications.

For more distant communication between vehicles and streetlights or buildings, known as car-to-infrastructure communications, photonics may still offer some advantages, Ducournau says, because stationary nodes tend to be connected to the rest of the communications system with optical fiber. Staying in the optical domain may simplify such connections. And with a data rate of 1 Tb/s, a vehicle wouldn't need to be in range of the infrastructure for very long; large volumes of data could be exchanged in short bursts when the car passes by a node on a lamppost.

In fact, the high data rate helps compensate not only for limited range but also for other potential shortcomings of terahertz frequencies. Because of the low power involved, it's not feasible to broadcast signals widely. Instead, it makes more sense to send narrow, targeted beams between transmitter and receiver. That requires using optical antennas to control the direction of the signals.

Directional antennas might seem to be a limitation, because a user could only point at one other user at a given time. But at 1 Tb/s, Jornet says, that's no problem. "If you communicate very fast, you don't need to be connected all the time," he says. "As long as you are connecting every now and then, when you're connected you just dump all your data or request all the data."

Jornet's group is building tiny, horn-shaped antennas with high sensitivity. Just a centimeter in size, the antenna provides gain of 30 decibels at 1 THz, increasing the signal a thousand-fold. By comparison, he says, an antenna providing similar gain at the WiFi frequency of 2.4 GHz would have to be half the size of a desk.

NASA, which uses terahertz radiation at higher frequencies to perform imaging in astronomical research, has built detector arrays for boosting signals. Such an approach might also work in automotive designs, he says. And another standard concept from optics, lenses that focus beams at those wavelengths, could also prove useful in controlling the direction of terahertz signals.

Tafur Monroy points out that directional signals also add an extra layer of security. If you're not broadcasting widely, there's less chance of on unwanted person intercepting your data.

SUBHED

In an effort to improve both the power output of terahertz generators and the sensitivity of detectors, Mona Jarrahi of the University of California, Los Angeles, turns to plasmonics. Plasmons are oscillations of electromagnetic energy. At visible wavelengths, they can arise at the point where metal meets air, but at terahertz frequencies they occur in semiconductors such as gallium arsenide. Jarrahi, a professor of electrical and computer engineering who runs the Terahertz Electronics Laboratory, uses a pair of near infrared lasers to create a beat signal at terahertz frequencies that she shines onto a semiconductor photodetector. "That is very easy to generate," she says. "I just need to make a very fast photodetector to detect these beats, generate a signal, and route it to an antenna to radiate those 1 THz beats."

She does this by building nanometer-sized structures on the detector, which concentrate the incoming light to the area immediately around them. These plasmonic nanostructures are connected to antennas, so when the light is absorbed by the semiconductor and creates charge carriers, those charge carriers have only a short trip to the antennas. "As soon as they're generated, they are routed to the output," she says. The setup increases both the output power and the sensitivity of detectors about a thousand-fold.

The use of terahertz frequencies is not limited to communications. Self-driving cars need to know where they are in relation to other vehicles, and terahertz broadcasts from other vehicles could be used to rapidly triangulate their location, much the way smartphones use signals from GPS satellites to find their location.

To take advantage of the frequencies' versatility, researchers in Finland have proposed dual-use systems that share some of the same equipment, such as steerable antenna arrays. "We could use those not only for high-powered, very rapid communication. We can also use them for positioning," says Markku Juntti, a professor of communications engineering at the University of Oulu in Finland. "It is very useful with smart traffic and the different autonomous systems."

Terahertz signals can also be used as a form of radar, as a substitute for or complement to cameras and lidar for imaging other vehicles, people, or obstacles. Their frequencies allow higher resolution than optical wavelengths, and they can see through fog better than lower-wavelength lasers.

Employing the same hardware for multiple uses would lower costs of self-driving cars, says Yevgeni Koucheryavy, a professor of electronics and communications engineering at Tampere University of Technology in Finland, who collaborates with Juntti. "It's clear that in vehicular applications we can benefit from simultaneous usage of terahertz waves to transmit information, but also to sense the environment." Juntti and Koucheryavy were joined in their work by researchers from Ericsson Research, part of the Swedish telecommunications company.

SEM photograph of a terahertz detector based on a logarithmic spiral antenna with plasmonic electrodes. Credit: Mona Jarrahi

The specifics of any of this technology will depend on the parameters-data rate, transmission distance, cost-that the designers of self-driving vehicles and smart highways require and what tradeoffs they're willing to accept. And there are a few years of development-more than five, fewer than ten, these researchers say-still to be done. "Do not think that you can just go and buy any of these things we're discussing today," Jornet says, although there are a few companies that make some terahertz components, mainly for military or research use.

There are other aspects of autonomous vehicles that need further development as well. Terahertz communications may be used in chip-level interconnects, to shunt data around the systems controlling the vehicle, but computer chips will still need processors and algorithms capable of handing the volume of data pouring in at ultrafast speeds. Photonics may play a role there as well, with optical devices to perform analog computations, taking some of the burden off digital systems and skipping the step of converting analog optical signals to digital.

At this intersection of radio and light waves, it remains an open question whether electronics or photonics will dominate. Jornet thinks the earliest systems will likely rely on photonics. "At optical frequencies, you are more used to handling very high bandwidth," he says. "It's easier to force down in frequency something that works than to move up in frequency." The specific systems developed will depend on which tradeoffs designers are willing to accept, but high-speed communication is coming soon, he says. "Will we be able to see one terabit per second within the next five years?" he asks. "For sure we will."

Neil Savage writes about science and technology in Lowell, Massachusetts.