With artificial intelligence and holograms create an impenetrable optical encryption system

Madrid. A new optical system that combines artificial intelligence (AI) with holograms to encode information has achieved an encryption level that traditional methods cannot penetrate.

This advance, according to researchers who have developed it, can pave the way for safer communication channels, helping to protect confidential data.

From rapidly evolving digital currencies to governance, medical care, communications and social networks, the demand for robust protection systems to combat digital fraud continues to growsaid in a statement the head of the Stelios Tzortzakis research team, of the Institute of Electronic and Laser Structure, Foundation for Hellas Research and Technology and the University of Crete, both in Greece.

Our new system achieves an exceptional encryption level when using a neuronal network to generate the deciphering key, which can only be created by the owner of the encryption system.

In the magazine OPTICALTzortzakis and his colleagues describe the new system, which uses neural networks to recover elaborately encoded and stored information as a hologram. They show that these trained can successfully decode the intricate spatial information in the coded images.

Our study provides a solid basis for many applications, especially cryptography and secure wireless optical communication, racing the path for next generation telecommunications technologiesTzortzakis said. “The method we develop is highly reliable even in difficult and unpredictable conditions, and addresses real problems – global disagreements such as adverse climatic conditions that often limit the performance of optical free space systems.”

Train neural networks

The researchers developed the new system after discovering that when holograms are used to encode a laser beam, it would be completely distorted randomly and the original lightning form could not be recognized or recovered by physical analysis or calculations. They acknowledged that this was an ideal way to encrypt information safely.

The challenge was to discover how to decipher the informationafirmó tzortzakis. We came up with the idea of ​​training neural networks to recognize the incredibly fine details of the coded light patterns. When creating billions of complex connections, or synapses, within neural networks, we could reconstruct the original forms of the light beam. This meant that we had a way to create the specific deciphering key for each encryption system configuration.

To create a physical system that mixes the rays of light completely and chaotic, the researchers used a high power laser that interacted with a small bucket full of ethanol. The liquid was not only economical, but also created the desired chaotic behavior in a short propagation distance of only a few millimeters. In addition to changing the intensity of the light beam that interacted with the liquid, it also exhibited thermal turbulence that strongly improved chaotic confusion.

Databases

To demonstrate the new method, experts applied it to encrypt and decode thousands of handwritten digits and other forms such as animals, tools and daily objects of well -established databases used as references to evaluate image recovery systems. After optimizing the experimental procedure and training the neuronal network, they demonstrated that it could accurately recover the encoded images between 90 and 95 percent of the time. They say that this rate could be even more improved with a broader training of the neuronal network.

Researchers plan to develop technology further by adding additional protection levels, such as two factors authentication. Since the greatest obstacle to the marketing of the system is the cost and size of the laser system, they are also investigating profitable alternatives to the high -power costly and bulky lasers.