Protecting Digital Privacy and Data Against the Threat of Future Quantum Computers
ECE Professor Yunsi Fei is leading a $600K NSF grant, in collaboration with Yingjie Lao from Tufts University, for “Secure and Efficient CRYSTALS-Kyber Architecture to Resist Against Side-channel and Fault Attacks on Post-Quantum Cryptography.” Quantum computers threaten to eventually break the cryptographic systems that protect everything from personal banking and healthcare records to national defense communications, making the migration to post-quantum cryptography (PQC) one of the most consequential infrastructure transitions of our time. Ensuring that this new generation of cryptography is securely deployed is essential to safeguarding the digital systems society will depend on for decades to come. This project aims to close a critical gap between efficiency and security in PQC ML-KEM implementations.
Abstract Source: NSF
The rapid advancement of quantum computing poses a serious threat to existing cryptographic systems, prompting the development of post-quantum cryptography (PQC) designed to resist the power of quantum computers. However, CRYSTALS-Kyber (Kyber), the scheme standardized by the National Institute of Standards and Technology (NIST) as the key encapsulation mechanism (KEM), remains vulnerable to side-channel attacks (SCAs) and fault attacks (FAs) that exploit physical leakages or induced faults to extract secret information. This project investigates the timely and important challenge of building secure PQC implementations and aims to close a critical research gap by developing secure Kyber architectures that can resist SCAs and FAs. This project identifies key opportunities to create strong countermeasures and combine them with hardware acceleration. This project also advances the hardware security and architecture of Kyber by exploring the interaction between algorithm design and hardware support, achieving both strong protection and high efficiency. This project will yield a novel framework for ensuring the security and efficiency of next-generation cryptosystems in the quantum era to meet the future needs of both commercial products and national defense. New findings will be disseminated broadly through presentations and tutorials at conferences and workshops.
This project pursues three closely connected research thrusts. Thrust 1 focuses on systematically evaluating side-channel leakage in Kyber, applying domain adaptation and transfer learning to study SCAs across different measurements, and designing realistic software-based SCAs. It also evaluates the side-channel resistance of individual computation modules to guide the development of effective countermeasures. Thrust 2 advances practical fault injection attacks by proposing a novel side-channel-assisted chosen ciphertext attack that removes the need for fault injections while still enabling secret retrieval. The team is also conducting realistic fault injections on Kyber using accessible equipment under more flexible fault models, surpassing prior simulation-based studies. Thrust 3 combines insights and countermeasures from the first two thrusts with hardware acceleration to deliver efficient and secure Kyber systems resilient to both SCAs and FAs.