SymTrust · Trust-by-Design – Strong Security Arguments for Symmetric Key Cryptography

Symmetric cryptographic primitives protect a large percentage of today's data, whether in transit or storage. This trend is going to increase even more in the future. Consequently, new domain-specific ciphers are needed with significant performance benefits for areas such as memory-encryption, multi-party computation, or crypto-currencies. Great progress has been made regarding their performance. Less progress has been made for security arguments that seek to proof their resilience against malignant attacks. Indeed, most of the security analysis we apply for symmetric ciphers is based on false assumptions and most attacks trying to break those ciphers are flawed. As a consequence, it takes several years to build the necessary trust in the security of any new symmetric cryptographic primitives to be deployed and used. In times of fast innovation cycles where emerging applications are in constant need for new tailored symmetric cryptographic solutions those slow design cycles of symmetric cryptographic primitives are prohibitive and have to be shortened. Within the project, I plan to enable stronger security arguments and tool-based bounds on key-recovery approaches which will accelerate the design process. This will open the path to a new design strategy of trustworthy by design ciphers and contribute to ready-to-use and highly-secure symmetric cryptography suitable to satisfy the needs of future applications.To this end, I will •analyze fundamental properties of symmetric cryptographic primitives especially related to their representation,•derive modular security arguments, •develop a software-assisted approach to optimize and verify the key-recovery part of attacks,•design an efficient cache-encryption algorithm to exemplify the potential of the new design framework. If successful, SymTrust will reshape the whole process of designing symmetric cryptographic ciphers to produce trustworthy by design primitives with the aim to shorten the time-to-deploy.