Laser security system protects online data

To better defend online data, a laser-based system to generate digital ‘keys’ can offer greater security than conventional approaches. The system, developed by KAUST, uses very little energy to produce a stream of complex keys, which act as unique fingerprints to safeguard data^[1].

“A key is a piece of information used to prove identity, protect data, or authorize access,” explains Yating Wan of the Integrated Photonic Laboratory at KAUST, who co-led the research team. “If the correct key is presented, the system accepts the user or device; if not, access is denied.”

Conventional digital security systems often rely on stored keys, passwords, or mathematically generated codes. These can be vulnerable if the stored information is copied, leaked, or reused.

To overcome that threat, some security systems produce a long series of different keys, each generated on demand and used only once. But generating truly random keys can be relatively slow, and often demands huge computing power.

Physical unclonable functions (PUFs) offer a potential solution. These systems exploit small physical differences in hardware to generate a unique key every time one is needed. “These differences naturally arise during fabrication and are extremely difficult to reproduce exactly,” explains Zhican Zhou, a Ph.D. student in Wan’s team. “A PUF adds a hardware-rooted layer of security because the secret is not simply stored in memory; it is dynamically generated from the physical behavior of the device itself.”

In collaboration with Boon Ooi’s group at KAUST, Wan’s team proposed using chaotic vertical-cavity surface-emitting lasers (VCSELs) as PUFs for dynamic authentication. These VCSELs are devices barely 200 micrometers wide, which can be fabricated on chips using commercial mass-production processes.

Each VCSEL produces a chaotic laser output with a varying intensity that is difficult to predict or reproduce, but which has unique statistical features. Varying the device’s drive current and operating temperature can produce further distinctive statistical features. By combining different devices with different operating states gives the PUF system a large set of distinct key-generating configurations. Each configuration’s chaotic response stream can be segmented into many dynamic keys that share the same configuration-specific statistical features.

The researchers then trained an AI model to recognize these features, so that it could check whether a key originated from one of the registered key-generating configurations, or whether it was a fake. They also developed an AI-assisted system to encode the key into a compact, hidden form for secure transmission.

To demonstrate their system, the researchers built an array of more than 100 VCSELs. Each device generates over 500 gigabits of data per second, enough to create a new key every 10 nanoseconds. A key contains more than 5,000 bits of entropy, roughly 20 times as much as today’s standard encryption keys.

The VCSELs have a very low energy consumption, requiring less than 1 picojoule (10^-12 J) per bit. Consequently, the system could generate trillions of keys using less energy than it takes to boil a cup of water.

The researchers now plan to build a prototype system that combines all of these elements into a compact security chip suitable for practical use.