Performance Evaluation of a Pseudo-Random Number Generator Against Various Attacks

Every user agrees upon a key-exchange protocol before initiating a communication over the Internet. These keys at the root level are a combination of highly unpredictable sequences of random numbers. It is crucial that the numbers generated are strong and secured. Ciphers are developed to generate these strong sets of keys. We conducted a thorough research on the existing hardware and software ciphers. Flaws were identified in the design complexity, platform dependency, application dimensions, and resource consumption. This led to the development of a generic n-bit LFSR based Cryptographically Secured Pseudo Random Number Generator (hardware-software co-simulation). As the generator forms the foundation for the security of the entire network, it has to be resistant to most attacks. The current ciphers which are used in real time applications like communication devices, online shopping and payments, group chats on social media, etc. are prone to attacks. This paper focuses on identifying those attacks and conducting the same on our cipher to evaluate its performance. To the best of our knowledge, we are the first to summarize these attacks from a wide domain of applications and prove it all experimentally in a single paper. Since a CSPRNG works at different layers of the communication cycle, attacks have been bifurcated into Cryptanalysis, LFSR, and Real-time attacks. Various test cases and results are discussed in order to ensure the effectiveness of the CSPRNG.