Fast Correlation Attacks through Reconstruction of Linear Polynomials

The task of a fast correlation attack is to efficiently restore the initial content of a linear feedback shift register in a stream cipher using a detected correlation with the output sequence. We show that by modeling this problem as the problem of learning a binary linear multivariate polynomial, algorithms for polynomial reconstruction with queries can be modified through some general techniques used in fast correlation attacks. The result is a new and efficient way of performing fast correlation attacks. Keywords. Stream ciphers, correlation attacks, learning theory, reconstruction of polynomials.

• Type: Conference
• Year: 2000
• Where: CRYPTO
• Authors: Thomas Johansson, Fredrik Jönsson
• Keywords: Correlation Attacks, CRYPTO 2000, Cryptology, Fast Correlation Attacks, Stream Ciphers