Abstract
Quantum computing threatens conventional public-key cryptography. In response, standards bodies such as NIST increasingly focus on post-quantum cryptography. In particular, hash-based signature schemes are notable candidates for deployment. No rigorous side-channel analysis of hash-based signature schemes has been conducted so far. This work bridges this gap. We analyse the stateful hash-based signature schemes XMSS and \(\text {XMSS}^{MT}\), which are currently undergoing standardisation at IETF, as well as SPHINCS—the only practical stateless hash-based scheme. While timing and simple power analysis attacks are unpromising, we show that the differential power analysis resistance of XMSS can be reduced to the differential power analysis resistance of the underlying pseudorandom number generator. This first systematic analysis helps to further increase confidence in XMSS, supporting current standardisation efforts. Furthermore, we show that at least a 32-bit chunk of the SPHINCS secret key can be recovered using a differential power analysis attack due to its stateless construction. We present novel differential power analyses on a SHA-2-based pseudorandom number generator for XMSS and a BLAKE-256-based pseudorandom function for SPHINCS-256 in the Hamming weight model. The first attack is not threatening current versions of XMSS, unless a customised pseudorandom number generator is used. The second one compromises the security of a hardware implementation of SPHINCS-256. Our analysis is supported by a power simulator implementation of SHA-2 for XMSS and a hardware implementation of BLAKE for SPHINCS. We also provide recommendations for XMSS implementers.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have