A Distinguisher for High-Rate McEliece Cryptosystems
Jean‐Charles Faugère, Valérie Gauthier–Umaña, Ayoub Otmani, Ludovic Perret, Jean–Pierre Tillich
Abstract
Open-access reader
Jean‐Charles Faugère, Valérie Gauthier–Umaña, Ayoub Otmani, Ludovic Perret, Jean–Pierre Tillich
Abstract
Open-access reader
The Goppa Code Distinguishing (GD) problem consists in distinguishing the matrix of a Goppa code from a random matrix. The hardness of this problem is an assumption to prove the security of code-based cryptographic primitives such as McEliece's cryptosystem. Up to now, it is widely believed that the GD problem is a hard decision problem. We present the first method allowing to distinguish alternant and Goppa codes over any field. Our technique can solve the GD problem in polynomial time provided that the codes have sufficiently large rates. The key ingredient is an algebraic characterization of the key-recovery problem. The idea is to consider the rank of a linear system which is obtained by linearizing a particular polynomial system describing a key-recovery attack. It appears that this dimension depends on the type of code considered. Explicit formulas derived from extensive experimentations for the rank are provided for “generic” random, alternant, and Goppa codes over any field. Finally, we give theoretical explanations of these formulas in the case of random codes, alternant codes over any field of characteristic two and binary Goppa codes.
OpenAlex reports 105 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
The Goppa Code Distinguishing (GD) problem consists in distinguishing the matrix of a Goppa code from a random matrix. The hardness of this problem is an assumption to prove the security of code-based cryptographic primitives such as McEliece's cryptosystem. Up to now, it is widely believed that the GD problem is a hard decision problem. We present the first method allowing to distinguish alternant and Goppa codes over any field. Our technique can solve the GD problem in polynomial time provided that the codes have sufficiently large rates. The key ingredient is an algebraic characterization of the key-recovery problem. The idea is to consider the rank of a linear system which is obtained by linearizing a particular polynomial system describing a key-recovery attack. It appears that this dimension depends on the type of code considered. Explicit formulas derived from extensive experimentations for the rank are provided for “generic” random, alternant, and Goppa codes over any field. Finally, we give theoretical explanations of these formulas in the case of random codes, alternant codes over any field of characteristic two and binary Goppa codes.
Key concepts: McEliece cryptosystem, Finite field, Cryptosystem, Mathematics, Cryptography, Linear code, Rank (graph theory), Discrete mathematics