Key to Crypto: Memristor Made by Inkjet Printer Unlocks True Random Number Generators
A group of researchers from KAUST (King Abdullah University of Science and Technology) New Breakthrough Manufacturing Technology Known as “Memristor” – A circuit that is one of the four basic electrical components along with resistors, capacitors, and inductors. This new technique has been shown to enable the creation of true random number generators (TRNGs), one of the key components of cryptography.
True random number generators are an important part of cryptography, and while perhaps counterintuitive (after all, how hard is it to generate random numbers?), even one of the most prone to failure I have. This is because random distributions (that is, where all possible events have equal probability of occurring) tend to be non-random distributions.
TRNGs are typically implemented at the silicon level, such as AMD’s Ryzen and Epyc Bound Cryptographic Coprocessors (CCPs) (currently in iteration 5.0). One of his methods of generating random numbers is by looking at phenomena that are inherently random, such as the photoelectric effect that underlies computer operation. These effects generate random numbers that serve as the basis for cryptographic operations. Each random number is transformed into part of the encrypted message in a process known as encipherment. hashing. To better understand this issue, think of AMD’s Xilinx division as a product. A field programmable gate array (FPGA) intended to act as a true random number generator.
However, electrical components have operating limits, and small voltage changes can cause computational or photoelectric “errors” that form patterns. Of course, if a pattern emerges in a pool of numbers that should be random, it’s really not random anymore. There is a pattern, with slightly different probabilities of some numbers being chosen over others. And if it’s not truly random, the pattern that emerges can be extracted, analyzed, and compared to the encrypted output… and the path to a message that is supposedly cryptographically secure can be found. is opened.
Some patterns can naturally emerge from certain imbalances in the system that push the system away from random “equilibrium” states (such as hardware degradation, where both the CPU and the GPU are slowing down over time). the maximum sustained operating frequency will be similarly reduced). ). We have seen these exploited by researchers. For example, data is extracted from patterns such as system fan speed. However, some can be compromised by sufficiently sophisticated attackers.
The work of KAUST researchers has enabled the fabrication of memristor-based TRNGs in a process similar to 3D printing. Instead of a regular filament, atomically thin layers of boron nitride and silver electrodes are deposited until the memristor elements are all stacked in place. This special manufacturing process makes the TRNG consume less power compared to typical CPU-integrated alternatives built with expensive circuits with millions of transistors (power usage and accelerator design costs both in terms of the space it occupies above).
“We used a novel two-dimensional layered material called hexagonal boron nitride to fabricate memristors, on which we printed silver electrodes using scalable and low-cost inkjet printing technology,” said the KAUS team. Researcher Pazos said. “The unique properties of 2D h-BN are preserved after electrode printing, allowing for excellent output and random signal generation.”
The resulting TRNG generator clearly matched the team’s expectations. We showed the best performance of TRNG in terms of random signal stability over time. We found the energy consumption to be incredibly low. Finally, an easy and fast circuit readout will allow her TRNG in memristor base to generate 7 million random bits per second for him.
“Furthermore, by interconnecting our memristor to a commercial microcontroller and performing live experiments with random number generation on-the-fly, we have demonstrated a built circuit that generates random numbers,” Pazos added.
And in contrast to most other technological breakthroughs, the technology looks ready for prime time out of the box. This technology can be easily deployed in Internet of Things (IoT) applications and other edge devices such as sensor node arrays.
“Our scalable, low-cost manufacturing method using inkjet printing not only enables superior performance, but is key to integrating these devices into low-cost, complex electronics,” Pazos said. . “This work demonstrates the potential of his 2D materials like h-BN to underpin a revolution in solid-state micro- and nanoelectronic devices and circuits due to their superior electronic, physical, chemical and thermal properties. doing.”