Wireless signals are getting smarter. Princeton engineers have unveiled a machine-learning system that allows ultrahigh frequency beams to curve around obstacles instead of being blocked by them.
Published August 18 in Nature Communications, the study shows how physics-informed AI can shape sub-terahertz “Airy beams” that deliver data speeds ten times faster than current networks, a breakthrough for virtual reality, autonomous vehicles, and next-generation communications.
How Curved Beams Beat Blockages
Signals at sub-terahertz frequencies (100–300 GHz) promise massive bandwidth, but they travel in narrow beams that objects easily block. Conventional fixes like reflectors are unreliable indoors. The Princeton team instead drew on a concept from 1979: Airy beams, waves that naturally curve in flight. By using transmitters that shape beams like curveballs, the system dodges obstacles without needing reflectors or a clear line of sight.
“This work tackles a long-standing problem that has prevented the adoption of such high frequencies in dynamic wireless communications to date,” said Yasaman Ghasempour, lead researcher and assistant professor of electrical and computer engineering at Princeton.
Training Beams Like Athletes
But bending a signal is only half the battle. There are infinite ways to curve a beam, and only a few maximize power at the receiver. To solve this, graduate student Haoze Chen and doctoral researcher Atsutse Kludze built a neural network trained on a custom simulator. Like NBA players perfecting shots through practice, the AI system learns when to send beams curving left, right, or over an obstacle.
- Sub-terahertz bands can carry 10× more data than today’s wireless.
- Airy beams allow waveforms to bend around walls or moving objects.
- Neural networks optimize curvature for real-time adaptability.
- Experiments showed stable 8K video streaming around obstructions.
Real-World Demonstrations
In lab experiments at 120 GHz, the researchers used custom metasurfaces to generate curved beams that streamed data even when a stone block blocked direct line of sight. The AI system adapted instantly as obstacles moved, cutting bit error rates by orders of magnitude compared to standard beams. The result: uninterrupted, high-quality transmission across complex and shifting environments.
Why It Matters
Sub-terahertz communication is considered the frontier of next-generation 6G networks. Beyond consumer uses like VR, the technology could strengthen communications for autonomous vehicles, rural broadband, and defense systems. As the study shows, physics-informed AI may be the key to making these fragile frequencies practical at scale.
The Future Of Curveball Signals
Wireless signals that think like athletes may sound futuristic, but Princeton’s work shows they’re real and scalable. If adopted, networks might soon throw their own curveballs, ensuring that immersive VR headsets, driverless cars, and next-gen defense systems never drop the ball. It is the beginning of a wireless era where losing connection to a wall is no longer acceptable.
Journal: Nature Communications. DOI: 10.1038/s41467-025-62443-0
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.