Security of implementations of Post-Quantum Cryptography algorithms

The security of the implementations of PQC algorithms is vital for maintaining the confidentiality, integrity, authenticity and availability of digital information and communications in the face of implementation attacks, such as, for example, side-channel attacks using information from timing, power consumption, electromagnetic radiation, fault attacks disturbing the secure of operation of the device and their combination. Such attacks, eventually also enhanced by the use of deep learning, constitute significant threats to both (embedded and regular) software and hardware implementations. In various application areas such as IoT, cloud-based applications, automotive, measures to prevent such attacks currently lead to substantial resource overhead due to the complexity of the algorithms, and the security remains unclear given the limited exploration of different attack surfaces. Countermeasures, to the extent that they are available, may have significant impact on run-time and memory consumption. Resistance in PQC implementations to implementation attacks is an increasingly common concern among customers, especially when exploring the right balance between security and performance.

Evaluating the security of PQC algorithm implementations against side-channel and fault attacks is crucial, given the proven vulnerabilities. Various countermeasures, such as masking, shuffling, randomized clocking, random delay insertion, constant weight encoding, code polymorphism, control-flow integrity and re-computation of critical operations can be employed to mitigate these attacks. Synergies between specific countermeasures and the design of cryptographic systems are available for pre-quantum cryptography but require investigation for post-quantum cryptography.

Proposals are welcome on developing solutions that protect against such implementation attacks, at reasonable costs and minimizing the loss of performance while maintaining the required security, as well as on the analysis of new attacks or combinations of attacks, also powered by the use of AI, for security-by-design approaches when designing Post Quantum Cryptographic systems. Activities can also lead to the development of testing methodologies and frameworks for automated security evaluations for correctness and resistance to remote side-channel attacks for regular software and for correctness and resistance to a broad range of implementation attacks for embedded software and hardware.