A framework is introduced for efficiently computing with encrypted data. We assume a semi-honest security model with two computing parties. Two different coding techniques are used with additively homomorphic encryption, such that many values can be put into one large encryption, and additions and multiplications can be performed on all values simultaneously. For more complicated operations such as comparisons and equality tests, bit-wise secret sharing is proposed as an additional technique that has a low computational and communication complexity, and which allows for precomputing. The framework is shown to significantly improve the computational complexity of state-of-the-art solutions on generic operations such as secure comparisons and secure set intersection.

Batching, Homomorphic encryption, Packing, Secure comparison, Secure equality, Secure set intersection, Vector addition chain
International Journal of Applied Cryptography

Veugen, P.J.M. (2020). Efficient coding for secure computing with additively-homomorphic encrypted data. International Journal of Applied Cryptography, 4(1), 1–15. doi:10.1504/IJACT.2020.107160