Research On Automatic Search Technology Of New Distinguishers For Block Cipher

Block cipher belongs to symmetric encryption system,which plays an important role in network and information security.Because cryptanalysis can more effectively promote the design innovation of block ciphers,so it has been a research hotspot in recent years.The main idea of statistical cryptanalysis is to look for the effective distinguisher of the algorithm and use the key recovery technology to obtain the full or partial key information of the algorithm.Generally,the number of distinguisher rounds is a more important factor for measuring the security of cryptographic algorithms.In this thesis,mixed integer linear programming?MILP?,automatic search technology,and key recovery technology are used to construct new integral distinguishers for the five latest block ciphers,and their security is estimated.The main research results are shown as follows:1.For ANU and ANU-II algorithms,some automatic search methods for new integral distinguishers of 9 and 8 rounds are proposed.respectively.According to the characteristics of round function,S-box and P-permutation of ANU cipher,a new MILP model is constructed base on division property,and the longest 9-round integral distinguisher is obttained.Similarly,The ANU-II algorithm is improved on the basis of ANU,the MILP model is constructed by using its structural characteristics,and the longest 8-round integral distinguisher is obtained.The research results show that the ANU-II algorithm has higher security,which provides an important reference for the evaluation of this type of algorithm.2.For Li Ci and ESF algorithms,some automatic search methods for new integral distinguishers of 12 and 9 rounds are proposed.respectively.Based on the specific features of Li Ci cipher,we deform and decompose its structure,and propose a new equivalent structure of Li Ci algorithm.The equivalent structure not only clarifies the analysis ideas,but also reduces the intermediate variables that involved when it performs MILP modeling.ESF algorithm is an improved version of LBlock.The algorithm adopts Feistel-SP structure network,and the round function is its core design.Bit-based on P-permutation is similar to PRESENT algorithm.According to the structural characteristics of the above two algorithms,a new MILP model is constructed based on division property,and reach a length of integral distinguishers of 12-round and 9-round of Li Ci and ESF algorithms are obtained.respectively.Compared with the existing results,the newly discovered distinguisher rounds are currently the best.3.The integral cryptanalysis based on division property of the 12 rounds of the GRANULE algorithm is given.By selecting a specific 8-round distinguisher and adding 4rounds backward,a key recovery attack scheme which is based on 12 rounds of GRANULE cipher is proposed.The required time complexity is about2^58.17,and the data complexity is about2^62.17 choices of plaintext.Compared with the known attack methods,this is the optimal analysis result for the complexity and the number of attack rounds.