New Methods for Symmetric Cryptography

Despite the worldwide adoption of cryptographic standards, the rise of Internet of Things creates a need for new cryptographic primitives tailored for resource-constrained environments. We focus on the design and implementation of lightweight linear layers for symmetric ciphers. We propose new constructions of lightweight MDS (Maximum Distance Separable) and near-MDS matrices. We present new cryptanalytic methods for two classes of emerging primitives: lightweight authenticated encryption schemes and ciphers dedicated to advanced protocols such as Multi-Party Computation and Fully Homomorphic Encryption. For the sake of implementation efficiency, many of the target ciphers utilize building blocks of low algebraic degree. To leverage this structural property, we enhance cube attacks with new degree evaluation and term enumeration techniques and revisit linear cryptanalysis by providing novel correlation computation methods. Moreover, automated tools are developed to search for distinguishers in the attacks. We also improve interpolation attacks in terms of memory complexity. The proposed methods enable us to present current best attacks on some ciphers including the first attack on full versions of MORUS. Encryption algorithms have shown to be crucial to address the security and privacy issues of data transformation on the Internet. However, they also prevent the work of intrusion detection based on deep packet inspection. We employ symmetric cryptographic primitives to design a privacy-friendly and market-compliant intrusion detection system over encrypted traffic. Our experiments show that our protocol is approaching feasibility in real-world applications.

Jaar van publicatie:2020

Toegankelijkheid:Open