Researchers at the Institute of Electronic Structure and Laser (IESL) at the FORTH research institute in Heraklion have combined holograms with Artificial Intelligence to create an uncrackable system of optical encryption. According to a news release from FORTH, the approach of optical encryption could provide increased security in the fields of crypto currencies, healthcare, communications and many other areas. “As the demand for digital security grows, researchers have developed a new optical system that uses holograms to encode information, creating a level of encryption that traditional methods cannot penetrate. This advance could pave the way for more secure communication channels, helping to protect sensitive data,” it says.
“From rapidly evolving digital currencies to governance, healthcare, communications and social networks, the demand for robust protection systems to combat digital fraud continues to grow,” says the lead researcher Professor Stelios Tzortzakis, of the IESL and the University of Crete. “Our new system achieves an exceptional level of encryption by utilising a neural network to generate the decryption key, which can only be created by the owner of the encryption system.”
Stelios Tzortzakis, Associate Professor at the Institute of Electronic Structure and Laser-FORTH. Prof. Tzortzakis received his Ph.D. from the École Polytechnique (France, 2001) in Nonlinear Optics. and has worked and collaborated with many research laboratories in Europe, Asia, and the USA. He is a recognised expert in nonlinear laser propagation phenomena. Photo: IESL-Forth.
In an article in Optica magazine, published by Optica Publishing Group, a Washington-based publisher of specialised journals on optics and photonics, Mr Tzortzakis and his colleagues describe the new system, which uses neural networks to reconstruct deeply encrypted information stored as holograms. They show that trained neural networks can successfully decode the complex spatial information in the encoded images.
“Our study provides a strong foundation for many applications, especially cryptography and secure wireless optical communication, paving the way for next-generation telecommunication technologies,” Mr Tzortzakis says. “The method we developed is highly reliable even in harsh and unpredictable conditions, addressing real-world challenges like tough weather that often limit the performance of free-space optical systems”.
Randomly scrambled light for security
The researchers developed the new system after discovering that when holograms are used to encode a laser beam, the beam becomes completely and randomly scrambled and the original shape of the beam cannot be recognised or recovered using physical analysis or computation. They realised that this was an ideal way of securely encoding information.
“The challenge was figuring out how to decrypt the information”, Mr Tzortzakis said. “We came up with the idea of training neural networks to recognise the incredibly fine details of the scrambled light patterns. By creating billions of complex connections, or synapses, within the neural networks, we were able to reconstruct the original light beam shapes. This meant we had a way to create the decryption key that was specific for each encryption system configuration”.
In order to create a physical system which would completely and chaotically scramble the beams of light, the researchers used a high-output laser interacting with a small cuvette filled with ethanol. The liquid was not only cheap, but exhibited the desired chaotic behaviour at a small propagation distance of a few millimetres. In addition to the change in the intensity of the light beam, the interaction of the light with the liquid also created thermal disturbances which greatly increased the chaotic encryption.
Successful encryption and decryption
To demonstrate the new method, the researchers used it to encrypt and decrypt thousands of handwritten figures and other shapes such as animals, tools and everyday objects, from established databases which are used as references for the assessment of image retrieval systems. After fine-tuning the experimental process and training the neural network, they showed that it could accurately recover the encrypted images in 90-95 per cent of cases. They say that this amount could be improved by further training of the neural network.
The researchers plan to develop the technology by adding further levels of protection such as two-factor authentication. Since the biggest obstacle to commercialising the system is the cost and size of the laser apparatus, they are also investigating economically viable alternatives to the bulky and expensive high-powered laser.
(Haniotika Nea, 01-02-25, www.forth.gr)