Processes and methods of selection of system-wide parameters of perspective algorithm of electronic signature based on algebraic lattices

Abstract

An important feature of the transition and post-quantum period is the application of new mathematical methods to counteract quantum cryptanalysis. The world cryptographic community pays special attention to the open competition for the post-quantum standard of electronic signature. The problem is to prove the stability of new mathematical methods for the synthesis of transformations such as electronic signature, in particular with the use of algebraic lattices. The existing algorithms of electronic signature of the 2nd stage of the NIST competition are analyzed. Among the selected candidates for the EP 2 standard, 3 of the 3 algorithms are based on algebraic lattices, CRYSTALS-DILITHIUM and FALCON. The NIST has issued a statement saying that it is most likely that one of the algorithms will be chosen due to the same mathematical basis used in both algorithms. The main attacks on electronic signature algorithms based on the problem of learning with errors, as well as the parameters of the EP Dilithium algorithm, which affect the stability and complexity of transformations, are considered. Methods for generating system-wide parameters of stability levels of 512 bits of classical and 256 bits of quantum security, as well as the protection of the algorithm againstfrom attacks by third-party channels are considered. The dependence of the time of electronic signature production on the keys is analyzed. The results of calculations for the level of stability 512/256 are given, and also recommendations on the choice of system-wide parameters are given. The results of the 2nd stage of the NIST competition of post quantum cryptographic algorithms, as well as the prospects of standardization of transformations such as electronic signature at the 3rd stage are considered. Conclusions are made about the need for a more detailed study of attacks on algorithms based on the problem of learning with errors, as well as the importance of generating system-wide parameters of higher levels.