Data confidentiality-preserving schemes for random linear network coding-capable networks

Abstract

Random linear network coding (RLNC) has been primarily devised as a throughput-efficient transmission scheme. However, due to its intrinsic security provided by the confusion resulting from the encoding operations, RLNC has also been proposed to be used in many security solutions to mitigate common security threats such as wiretap attacks. In this paper, we propose two encryption schemes as a solution to the wiretap problem in an RLNC-capable network. Both of our encryption schemes rely on securing the encoding coefficient matrix as well as the partial permutation of the data matrix symbols after the application of the T transformation. This latter is used to represent the data matrix over a smaller finite field to increase the number of possible permutations as well as the confusion and diffusion properties of the system. Comparative analysis shows that our schemes are more computationally secure and provide better confusion and diffusion than the Secure Practical Network Coding (SPOC) and P-Coding, which are state-of-the-art schemes. The execution times of both schemes stand between those of P-Coding and SPOC, which verifies their lightweight nature.