Construction of a Cryptographically Secure Pseudorandom Sequence Generator Based on the Blender Algorithm

The paper presents a scheme of a cryptographically secure pseudorandom sequence generator. The generator consists of three blocks that provide random data collection and processing, the internal state preparation and pseudorandom sequences generation. The main algorithm for all blocks is a blender from the 2-EXT extractor. For the first block, authors developed a special bit mixing algorithm that increases the min-entropy of raw data. The mixing procedure and the following processing with the help of a blender have shown high effectiveness. The generator software implementation results were analyzed using the NIST 800-90 and NIST 800-22 tests, as well as Pearson criterion. The study showed that the output sequences of the constructed generator have a high level of min-entropy, good statistical properties and a distribution close to uniform. The comparative analysis confirmed that the proposed generator is similar in quality of output sequences with generators based on “Kuznechik” and AES ciphers. However, application of strong blender and the special procedure of forming and using keys increase its cryptographic strength.