Also referred to as the Holy Grail of Cryptography, Fully Homomorphic Encryption (FHE) allows for arbitrary calculations over encrypted data. As a basic use-case, FHE enables a User to delegate a computation over her sensitive data to a semi-trusted Cloud: in such a setup, the User provides her data encrypted to the Cloud, the Cloud evaluates a function over the encrypted data without ever decrypting it, and sends the (encrypted) result back to the User, who finally decrypts it and obtains the desired value . However, even after more than twelve years of advances in this field, FHE schemes are still fairly slow in evaluation, therefore any optimization is welcome. Among existing FHE schemes, in this contribution we focus on the TFHE Scheme by Chillotti et al., which currently achieves the best evaluation times for generic functions. To be instantiated, TFHE however requires an extensive set of parameters. These parameters affect several aspects, including but not limited to the cleartext size, the bit-security level, the probability of errors and also the evaluation time. We propose, implement and evaluate a (semi-)automated approach to generate a set of TFHE parameters with particular respect to the evaluation time, given just the desired security level, cleartext precision, and a parameter that relates to the properties of the evaluated function . With our tool, we re-generate some of the existing TFHE parameters, while achieving up to 39% better execution times in an equivalent setup.
