Meerpartijen blokcijfers en toepassingen

This thesis investigates the evaluation of block ciphers using secure multi-party computation (MPC). A block cipher is a symmetric-key primitive that encrypts a fixed-length input block under a secret key and outputs a block of the same length. MPC protocols allow a set of mutually distrusting parties to jointly compute a function on their private input while the protocol execution does not reveal any information about other parties' private inputs. The potential of MPC to enable privacy-preserving computation has attracted a significant amount of academic research, considerably advancing efficiency. Evaluating symmetric-key constructions in MPC leads to several important applications, including threshold key management and conversions between encrypted and secret-shared data. In this thesis, we propose novel MPC protocols to efficiently compute substitution-permutation network (SPN) block ciphers in MPC, where both the secret key and the input block are secret-shared. Obtaining efficient protocols to compute traditional SPN ciphers, such as the Advanced Encryption Standard (AES), is challenging due to their non-linearity, iterative structure and CPU-centric design mixing Boolean and arithmetic steps. In contrast, many MPC protocols work on mathematical structures, such as fields or rings, and incur communication cost for non-linear operations. This thesis proposes novel protocols with the following contributions. (1) We lower the multiplicative complexity of the function representation of S-boxes within SPN ciphers and thus reduce communication and local computation cost when executed using MPC protocols. (2) We lower the communication round complexity for non-linear layers. Many contributed techniques are general and can be adapted to other SPN ciphers. As an application to the secure computation of block ciphers, this thesis also explores block cipher-based encryption schemes and presents an MPC-friendly mode of operation for forkciphers that is round-optimal for decryption.