Hardware Architectures for Post-Quantum Cryptography

Wen Wang, Yale University Graduate School of Arts and Sciences


The rapid development of quantum computers poses severe threats to many commonly-used cryptographic algorithms that are embedded in different hardware devices to ensure the security and privacy of data and communication. Seeking for new solutions that are potentially resistant against attacks from quantum computers, a new research field called Post-Quantum Cryptography (PQC) has emerged, that is, cryptosystems deployed in classical computers conjectured to be secure against attacks utilizing large-scale quantum computers. In order to secure data during storage or communication, and many other applications in the future, this dissertation focuses on the design, implementation, and evaluation of efficient PQC schemes in hardware. Four PQC algorithms, each from a different family, are studied in this dissertation. The first hardware architecture presented in this dissertation is focused on the code-based scheme Classic McEliece. The research presented in this dissertation is the first that builds the hardware architecture for the Classic McEliece cryptosystem. This research successfully demonstrated that complex code-based PQC algorithm can be run efficiently on hardware. Furthermore, this dissertation shows that implementation of this scheme on hardware can be easily tuned to different configurations by implementing support for flexible choices of security parameters as well as configurable hardware performance parameters. The successful prototype of the Classic McEliece scheme on hardware increased confidence in this scheme, and helped Classic McEliece to get recognized as one of seven finalists in the third round of the NIST PQC standardization process. While Classic McEliece serves as a ready-to-use candidate for many high-end applications, PQC solutions are also needed for low-end embedded devices. Embedded devices play an important role in our daily life. Despite their typically constrained resources, these devices require strong security measures to protect them against cyber attacks. Towards securing this type of devices, the second research presented in this dissertation focuses on the hash-based digital signature scheme XMSS. This research is the first that explores and presents practical hardware based XMSS solution for low-end embedded devices. In the design of XMSS hardware, a heterogenous software-hardware co-design approach was adopted, which combined the flexibility of the soft core with the acceleration from the hard core. The practicability and efficiency of the XMSS software-hardware co-design is further demonstrated by providing a hardware prototype on an open-source RISC-V based System-on-a-Chip (SoC) platform. The third research direction covered in this dissertation focuses on lattice-based cryptography, which represents one of the most promising and popular alternatives to today's widely adopted public key solutions. Prior research has presented hardware designs targeting the computing blocks that are necessary for the implementation of lattice-based systems. However, a recurrent issue in most existing designs is that these hardware designs are not fully scalable or parameterized, hence limited to specific cryptographic primitives and security parameter sets. The research presented in this dissertation is the first that develops hardware accelerators that are designed to be fully parameterized to support different lattice-based schemes and parameters. Further, these accelerators are utilized to realize the first software-harware co-design of provably-secure instances of qTESLA, which is a lattice-based digital signature scheme. This dissertation demonstrates that even demanding, provably-secure schemes can be realized efficiently with proper use of software-hardware co-design. The final research presented in this dissertation is focused on the isogeny-based scheme SIKE, which recently made it to the final round of the PQC standardization process. This research shows that hardware accelerators can be designed to offload compute-intensive elliptic curve and isogeny computations to hardware in a versatile fashion. These hardware accelerators are designed to be fully parameterized to support different security parameter sets of SIKE as well as flexible hardware configurations targeting different user applications. This research is the first that presents versatile hardware accelerators for SIKE that can be mapped efficiently to both FPGA and ASIC platforms. Based on these accelerators, an efficient software-hardwareco-design is constructed for speeding up SIKE. In the end, this dissertation demonstrates that, despite being embedded with expensive arithmetic, the isogeny-based SIKE scheme can be run efficiently by exploiting specialized hardware. These four research directions combined demonstrate the practicability of building efficient hardware architectures for complex PQC algorithms. The exploration of efficient PQC solutions for different hardware platforms will eventually help migrate high-end servers and low-end embedded devices towards the post-quantum era.