Power/Performance Bits: Oct. 19

Post-quantum crypto chip; dissolving electronics.


Post-quantum crypto chip
Researchers at the Technical University of Munich (TUM) designed and had fabricated an ASIC to run new encryption algorithms that can stand up to quantum computing.

“Ours is the first chip for post-quantum cryptography to be based entirely on a hardware/software co-design approach,” said Georg Sigl, Professor of Security in Information Technology at TUM. “As a result, it is around 10 times as fast when encrypting with Kyber – one of the most promising candidates for post-quantum cryptography – as compared to chips based entirely on software solutions. It also uses around eight times less energy and is almost as flexible.”

The team modified an open-source chip design based on RISC-V. The chip’s post-quantum cryptography capabilities are facilitated by a modification of the processor core and special instructions that speed up the necessary arithmetic operations.

The design also incorporates a purpose-designed hardware accelerator. As well as supporting lattice-based post-quantum cryptography algorithms such as Kyber, it could also work with the SIKE algorithm, which requires much more computing power. According to the team, the chip developed at TUM could implement SIKE 21 times faster than chips using only software-based encryption. SIKE is seen as a promising alternative if the time comes when lattice-based approaches are no longer secure.

Additionally, the team intentionally built hardware trojans into the chip to experiment with ways of detecting them.

“We still know very little about how hardware trojans are used by real attackers,” said Sigl. “To develop protective measures, we need to think like an attacker and try to develop and conceal our own trojans. In our post-quantum chip we have therefore developed and installed four hardware trojans, each of which works in an entirely different way.”

The researchers plan to test the chip’s cryptography capabilities and the detectability of the hardware trojans. They will also incrementally shave off the chip’s circuit pathways while photographing each successive layer. The goal is to try out new machine learning methods for reconstructing the precise functions of chips even when no documentation is available.

“These reconstructions can help to detect chip components that perform functions unrelated to the chip’s actual tasks and which may have been smuggled into the design,” said Sigl. “Processes like ours could become the standard for taking random samples in large orders of chips.  Combined with effective post-quantum cryptography, this could help us to make hardware more secure – in industrial facilities as well as in cars.”

Dissolving electronics
Researchers from Tianjin University and Tsinghua University developed a material that can be used in dissolvable electronics.

The two-metal nanocomposite was used in a prototype smartwatch that dissolved when submerged in water for 40 hours. Based on zinc with silver nanowires added to make it more conductive, the metallic solution was screen-printed onto pieces of poly(vinyl alcohol), a polymer that degrades in water. The circuits were solidified by applying small droplets of water that facilitate chemical reactions and then evaporate.

With this approach, the team made a smartwatch with multiple nanocomposite-printed circuit boards inside a 3D printed poly(vinyl alcohol) case. The smartwatch had sensors that accurately measured a person’s heart rate, blood oxygen levels and step count, and sent the information to a cellphone app via a Bluetooth connection.

The outer package held up to sweat, but once the whole device was fully immersed in water, both the polymer case and circuits dissolved completely within 40 hours. All that was left behind were the watch’s components, such as an organic light-emitting diode (OLED) screen and microcontroller, as well as resistors and capacitors that had been integrated into the circuits.

The researchers said the two-metal nanocomposite can be used to produce transient devices with performance matching that of commercial models.

Leave a Reply

(Note: This name will be displayed publicly)