24/06/2021 - 09:30 Gaspard Anthoine and Charlotte Lefevre (UGA) IMAG 106 Aussi accessible via https://meet-ljk.imag.fr/b/pie-sxg-kmz-6fm
We will have two talks from master's students who did their internship within the team. Gaspard Anthoine: Secure cloud-based outsourcing for data externalization In the context of remote storage on insecure resources, proofs of retrievability provide the user with guarantees that the remote server owns the data. The internship focuses on improving client computations for a new protocol that solves this problem efficiently, in practice by relying on probabilistic certification techniques in linear algebra. We will thus study efficient techniques for verifiable computations on encrypted data when the data can be updated. For this it will be necessary to mix techniques from verifiable computing, proven retrievability or succinct non interactive proofs of knowledge. Charlotte Lefevre: Time-Memory trade-offs for large-weight syndrome decoding in ternary codes An important class of post-quantum-resistant algorithms is based on linear codes. Among them Wave is a new signature scheme based on linear codes, whose generic security partly relies on the syndrome decoding problem over GF(3) with large weight. Wave's choice of parameters is based on the current asymptotically-fastest attack, which has equal time and memory cost. However, practical attacks on cryptosystems have seldom a memory cost equal to the time cost, so Wave's security estimation is unrealistic. In this talk we will present an information-set decoding framework, and instantiate it with several birthday-based algorithms based on the k-tree algorithm and the dissection framework. This provides an interesting range of time-memory trade-offs of the form T = M^i with i > 1. These results are useful to have a finer estimation of Wave's security.