Title:Generalized rank weights of reducible codes, optimal cases and related properties

Abstract:Reducible codes for the rank metric were introduced for cryptographic purposes. They have fast encoding and decoding algorithms, include maximum rank distance (MRD) codes when Gabidulin codes may not be applied and can correct many rank errors beyond half of their minimum rank distance, which make them suitable for network coding. Particular cases constituted by cartesian products of MRD codes have already been proposed for secure network coding. In this paper, we give lower and upper bounds on generalized rank weights (GRWs) of general reducible codes, which measure information leakage on the network when used in the form of coset coding. We find new parameters for which these codes are MRD or close to MRD, while having MRD components at the same time. Afterwards we obtain all linear codes with optimal GRWs for all possible fixed packet and code sizes, up to rank equivalence. We see that all of these optimal codes have moreover efficient decoding algorithms using any of their bases. Then we see that the security performance of general reducible codes is better than the threshold values given by their GRWs. Finally, we study some related properties: We give conditions for reducible codes to be rank equivalent to cartesian products of linear codes and conditions to be rank degenerate. We study their duality properties and MRD ranks, and give new alternative constructions to the classical (u, u+v) construction. A small remark on consequences of the given study on the GPT cryptosystem is given in the appendix.