When you send a message on WhatsApp, iMessage, or Signal, no one but you and the receiver can read the text. Any eavesdropper tapping in to the data sent between phones would just see a random, incomprehensible string of characters, and only the recipient has the key needed to decipher that string back into readable text.

But can one still convey secret information if phone networks are down?

A new paper in Advanced Optical Materials suggests it is possible. While it is still an experimental technique, the researchers hope their proposal might pave the way towards more secure anti-counterfeiting technologies for passports, IDs or bank notes, and even towards tools to aid disaster response in the event of a network failure.

Putting the pieces together

The paper focuses on printing images in a way that they can only be recovered by an intended recipient. The protocol starts with a greyscale image whose pixels are first distributed across different channels of the infrared spectrum.

Much like computers store each pixel of a color image in red, green and blue channels (making up the RGB code for each color), one can store different chunks of a greyscale image in different infrared wavelengths — potentially dozens of different ones, although the current experiment uses only four. If one were to put the different wavelengths back together, the original image could be recovered.

Next, the pieces corresponding to each channel are scrambled in a prescribed way, such that, if they were put back together, nothing recognizable would come out — just a random array of pixels. This scrambled result is then recorded onto a special material, called a metamaterial, that can store different infrared wavelengths and keep them separate thanks to wafer-thin layers of material forming very particular 3D structures.

The point of the scrambling process is that it is reversible — as long as one knows how it was carried out in the first place. The recipe is hidden in a secret key which should only be shared with the intended recipient. But now, by using a special camera capable of reading the different infrared wavelengths, that recipient can unscramble the result and recover the original image, while an eavesdropper cannot.

A bigger future

The experiment performed for the study involved took place at a microscopic scale. However, Michelle Povinelli, senior author of the study and professor of electrical engineering and physics and astronomy at the University of Southern California, imagines a future where these encrypted images can be made much bigger, at a scale perceivable by the human eye.

Bigger images “could be useful for humanitarian and disaster recovery efforts,” she said. If a natural disaster compromises the phone network, it might be useful to label the location of crowbars, dynamite, or even bodies securely, so that responders can access them but looters cannot.

Povinelli suggested that an encrypted image label could be stuck to the wall of a building, and only responders with specialized equipment and, importantly, access to the encryption key could read the information it contains.

Another possible future application is anti-counterfeiting tags like the silver stripes or the holograms present in IDs, passports and bank notes. “Authorized people who are given the key [could] check” they are authentic, Povinelli said, since they would be able to see the encoded image correctly while a fake document would fail to display it.

Upping the barrier

These new tags would also be harder to duplicate than the current ones. “Not many people are able to make a hologram such that when you rotate it or illuminate it in a certain way, a logo comes out,” said Enrique Tajahuerce, professor of optics at the Universitat Jaume I in Castellón, Spain, who was not involved in the study.

But, since anyone can see the hidden image in current holograms, forgers at least know what they are aiming for. With this new technique, instead, forgers would need a lot more fidelity in copying the microscopic layers of the metamaterial, since they would not know what image they are trying to convey. “We’re upping the barrier to counterfeiting,” Povinelli said.

Before these applications can become a reality, some practical issues will need to be addressed. The scrambling process being reversible means that the same key is used to encrypt and decrypt the image — this is known as “symmetric encryption”. Therefore, whoever can read the image also has the power to encrypt new images, so it is very important that the key is shared in a secure manner to ensure that it only reaches intended recipients.

A cat-and-mouse game

Further, the encryption scheme will need to stand the test of time to make sure it is actually secure. When a new scheme is proposed, it is standard to publish the details and invite the research community to try and hack it.

If it remains safe after a long enough period, it is deemed adequate for real applications. “It’s a cat-and-mouse game,” said Artur Carnicer, professor of optics at the University of Barcelona who was not involved in the study.

Other concerns involve what Povinelli called “practical engineering work”. This would address the durability of a tag on a bank note that gets passed around hundreds of hands and potentially worn due to rubbing, or ambient light and other sources of noise that might get in the way of an adequate reading of the tag on a building after an earthquake. Coating the tags to make them more durable or including some calibration method to deal with noise could be possible fixes to these problems.

Thus, the current paper provides a proof of concept that these ideas might, one day, become a reality. The basic technology is already available, so the wait might not even be that long. “We could see it in the next few years,” said Tajahuerce.

Reference: Romil Audhkhasi, et al., Experimental Implementation of Metasurfaces for Secure Multi-channel Image Encryption in the Infrared, Advanced Optical Materials (2023). DOI: 10.1002/adom.202203155

Feature image credit: Sigmund on Unsplash