The researchers used a specially designed metasurface to direct the transmissions.
Ultra-fast wireless signals have a weakness: they can’t see walls. As engineers push into the sub-terahertz spectrum to handle the enormous data demands of virtual reality and autonomous vehicles, even a bookcase or a passing person can block a signal, causing data loss.
Now, researchers at Princeton University have developed a system that allows these high-frequency signals to bend around obstacles, keeping data flowing in even the most cluttered environments.
The breakthrough uses a combination of physics and machine learning to create “Airy beams”—curved transmission paths that can navigate around objects instead of bouncing off them.
First proposed in 1979, Airy beams had mainly been studied for their physics. Princeton’s team went a step further: they trained a neural network to select the optimal beam in real time for any environment, adapting as obstacles appear or move.
Lead researcher Yasaman Ghasempour, an assistant professor of electrical and computer engineering at Princeton, said, “As our world becomes more connected and data-hungry, the demand for wireless bandwidth is soaring.
Sub-terahertz frequencies open the door to far greater speeds and capacity.” She added that the work is an important step toward deploying data transmission in the sub-terahertz band, which could handle 10 times the data of current wireless systems.
Unlike lower-frequency radio waves that spread widely, sub-terahertz signals travel in tightly focused beams, making them vulnerable indoors.
Previous solutions relied on reflectors to bounce signals around obstacles, but these are not practical in most real-world settings. The new approach allows the signal itself to curve, much like a curveball in baseball, using precise beam shaping.
Graduate student Haoze Chen, the paper’s lead author, said, “This is for complex indoor scenarios where you don’t have line of sight. You want the link to adapt to that.”
He added, “Most work with Airy beams has focused on creating the beams and exploring their underlying physics. What we are doing is not only generating the beams but finding which beams work best in the situation. People have shown that these beams can be created, but they have not shown how the beams can be optimized.”
To train the neural network, the team developed a simulator that could model countless indoor scenarios without physically testing each one.
Chen explained, “For Airy beams, this is impractical. There are infinite ways of curving, depending on the degree of the curve and where the curve happens. There is no way a transmitter can scan through.”
Coauthor Atsutse Kludze said, “Throwing a lot of data at the neural net is not effective. Instead, we use principles from physics to create and train the neural net.”
Once trained, the system adapts incredibly quickly, maintaining strong connections even in crowded, constantly changing environments.
The team tested the system in experimental setups designed to mimic complex, real-world indoor environments. While the experiments focused on understanding and controlling the technology, the results suggest practical applications are within reach.
These include ultra-fast VR systems, fully autonomous vehicles, and future indoor wireless networks that can transmit massive amounts of data without interruption.
Ghasempour said, “This work tackles a long-standing problem that has prevented the adoption of such high frequencies in dynamic wireless communications to date. With further advances, we envision transmitters that can intelligently navigate even the most complex environments, bringing ultra-fast, reliable wireless connectivity to applications that today seem out of reach—from immersive virtual reality to fully autonomous transportation.”
The findings have been published in the journal Nature Communication.
MasterCard