Integer Linear Programming for Three-Subset Meet-in-the-Middle Attacks: Application to GIFT

Abstract

This article presents a new usage of integer-linear-programming (ILP) for block-cipher analysis, in particular for automating a procedure to search for optimal independent key bits used in a meet-in-the-middle (MitM) attack. The research is motivated by a recent lightweight block-cipher design GIFT, in which the evaluation by the designers has some room to be improved. The developed tool finds optimal choices of independent key bits, which improves the complexity of the 15-round MitM attack, the current best attack, on GIFT-64 from 2 120 to 2 112.