近日，中国科学院大学密码学院在对称密码分析领域取得最新研究进展。密码学院特别研究助理刘勇作为第一作者撰写的论文《Differential-based Attacks on Round-reduced BAKSHEESH》被The 29th Information Security Conference（ISC 2026）接收。
该研究重点分析了 BAKSHEESH 算法抵抗差分、差分-线性和 Boomerang 攻击的能力，取得了针对该算法的一系列新的差分类攻击成果。其中，所提出的差分攻击将已有最佳差分攻击的轮数提高了一轮。
ISC 2026 于2026年10月28日至30日在法国雷恩举行，由 IRISA 实验室的 SUSHI 研究组（the SUSHI research group at the IRISA laboratory）组织。2026年10月，密码学院特别研究助理刘勇代表研究团队参加会议，并在大会上作学术报告，介绍该项研究成果。
Abstract. BAKSHEESH is an efficient lightweight block cipher that features both GIFT-like and ZORRO-like structures. In this paper, we first construct Mixed-Integer Linear Programming models that constrain the Hamming weights of the input and output differences/masks of the middle part, enabling the efficient search for advantageous differential-linear (DL) and boomerang distinguishers under limited computational resources. Furthermore, by exploiting decision tree techniques and the two-round differential propagation property of BAKSHEESH, we parameterize the number of guessed key bits and the logarithm of the number of possible plaintext/ciphertext differences, and subsequently build MILP models to identify the best DL and boomerang key-recovery attacks under the stated constraints. As a result, we identify the first 17-round DL distinguisher and 16-round boomerang distinguisher for BAKSHEESH, yielding the first 19-round DL attack and 18-round boomerang attack with time complexities of 2^{126.97} and 2^{126.48}, respectively. In addition, we construct the first 20-round chosen-ciphertext differential attack with a time complexity of 2^{122.59}, improving the best previously known differential attack by one round. Overall, our results provide new insights into the security of BAKSHEESH against differential-based cryptanalysis.
作者信息：Yong Liu, Siwei Sun∗, Zejun Xiang, Rui Xu, Xiutao Feng, and Xiangyong Zeng: “Differential-based Attacks on Round-reduced BAKSHEESH,” in The 29th Information Security Conference（ISC 2026）
