The invention of radio communications in the 1890s was a discovery that changed the world.
A signal could be coded into an electromagnetic radio wave, sent forth into the world, and picked up by a receiver, which translates and converts the signal back to an understandable format.
With current technology, the electromagnetic signal is typically converted into an electrical signal and digitized into numbers, which can then be processed to extract the information it carries.
Now, a team of researchers led by Southeast University in China has shown that some of those calculations can instead be performed directly on the radio waves as they interact with a programmable surface.
“In an everyday analogy, it can be viewed as a reprogrammable electromagnetic-space computer to process the signals directly in the electromagnetic wave propagation,” electrical engineers Jun Yan Dai, Qiang Cheng, and Tie Jun Cui, the study’s corresponding authors, told ScienceAlert.
“The computed results are carried out directly by the outgoing electromagnetic wave, without requiring to firstly convert the entire signal into digital data for processing by a conventional processor.”
Radio waves occupy the low-frequency end of the electromagnetic spectrum – basically, light – and have a set of properties that make them especially useful for transmitting information.
They can travel long distances through the atmosphere, diffract around obstacles, travel through some materials, and many can travel through weather that interferes with shorter wavelengths. We use them for all sorts of technology, from Wi-Fi to radar to GPS to broadcast media.
But transmitting a radio wave is only part of the job. For many applications, we also need to precisely manipulate those waves – changing properties such as their direction, phase, or amplitude.
One increasingly powerful way to do that is with a metasurface – a thin, engineered surface covered in tiny structures designed to alter electromagnetic waves in carefully controlled ways.
The metasurface developed by the research team takes that idea a step further. Its properties can be electronically reprogrammed, allowing the researchers to change how an incoming radio wave is manipulated over both space and time.
“This architecture builds on the long-term development of time-coding metasurfaces. In 2018, our group introduced the concept of time-domain digital coding metasurfaces, extending the metasurface control from the static spatial domain into the time domain,” the researchers explained.
“Building on this foundation, we established a direct relationship between time-coding sequences and mathematical operations such as Fourier transforms and convolutions. This allowed the metasurface to move beyond controlling electromagnetic waves and to use electromagnetic waves as a computational medium.”
They chose the two operations mentioned – Fourier transforms and convolutions – because they represent two of the most fundamental operations in signal processing.
A Fourier transform is an operation that breaks a pattern up into its parts – think of separating a chord into its constituent notes, or a white beam of light into a rainbow. That’s the simplest explanation, but the math involved is more complicated.
frameborder=”0″ allow=”accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share” referrerpolicy=”strict-origin-when-cross-origin” allowfullscreen>Convolutions, on the other hand, can be used to find where a known pattern occurs within a signal. For example, in radar, finding where a reflected pulse matches the transmitted one can help determine how far away an object is.
Both operations have practical applications in radar. Fourier transforms can reveal the Doppler shift in a reflected signal – how much the wavelength is stretched or squished – to see how fast an object is moving. Convolutions, as mentioned, can determine distance.
But the researchers didn’t build two separate metasurfaces. Instead, they made one that can alternate between the two tasks by switching the time-coding sequence used to program the surface.
“We program the metasurface with a time-coding sequence that represents the required computational kernel. When the radio-frequency signal reaches the metasurface, it is directly modulated by this time-varying response,” the researchers said.
“The resulting wave components then combine coherently, producing the Fourier transform or convolution results in the outgoing electromagnetic wave. In this sense, the essential multiplication-and-summation process is performed physically in one time rather than calculated point by point by a digital processor.”
Basically, the surface is programmed with a changing pattern that corresponds to the desired mathematical operation. The incoming radio wave hits it and is reshaped by the metasurface in a carefully controlled way. Different components of the resulting wave combine to carry out the calculations.
The researchers tested the metasurface in different settings – inside a clean lab as well as outdoors with realistic multipath and clutter conditions – and found that it was accurately able to perform both operations in real-time.
For example, they fed the system a simulated radar echo from a target moving at 200 meters (656 feet) per second. It returned a velocity of 201 meters per second with 500 sampling points, and 199.5 meters per second with 1,000 sampling points – both within the system’s resolution.
frameborder=”0″ allow=”accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share” referrerpolicy=”strict-origin-when-cross-origin” allowfullscreen>They also tested its ability to measure distance with simulated targets from 4 to 28 kilometers (2.5 to 17.5 miles) away. The errors remained within the system’s 300-meter resolution.
And all of this could be done by switching the metasurface from Fourier transform to convolution without modifying the hardware.
That doesn’t mean the conventional hardware is no longer required; as the researchers noted: “Conventional electronics are still used to generate and load the time-coding sequences, drive and synchronize the metasurface, and receive and display the output.”
However, it could result in a simpler radio-frequency-signal processing chain that could reduce reliance on components such as high-speed analog-to-digital converters and digital processors.
The next hurdle will be to increase the modulation speed of the metasurface while maintaining accuracy, stability, and energy efficiency. Faster modulation is needed to process wider-band signals, but it can bring greater signal losses and more complex hardware.
When they get there, the obvious first application will be radar. From there, however, it may be possible to implement the system across a range of radio applications, including 6G communications, satellite systems, and automotive radar.
Related: Mystery of Weird Signal Detected in The Australian Desert Finally Solved
“Ultimately,” Dai, Cheng, and Cui told ScienceAlert, “we envision a single hardware platform that can switch flexibly between different computational tasks simply by changing its time-coding sequence.”
The findings have been published in Nature Communications.
This article was fact-checked by Clare Watson and edited by Clare Watson. While we pride ourselves on our process, we are only human. If you spot a mistake, please let us know.
