Security Analyses of Different LFSR-Based Ciphers to Propose a Novel Approach Compatible with Parallel Computing Platform, Providing Resistance Against Various LFSR-Based Attacks
Development of a framework, for generating sets of random numbers which are highly nondeterministic and the dimensional distribution of which is strong, is need of the hour. Applications of pseudorandom numbers are widespread in areas of keying, re-keying, authentication, smart phone security, etc. Their use is central in the network security domain. Hence, developing a cryptographically secure pseudo-random number generator (CSPRNG) would be beneficial. The proposed generic model is constrained to linear feedback shift registers (LFSR), owing to its good statistical properties, large period, well suited to low power or high speed requirements. The use of pseudo-random numbers are high in hardware areas like wireless devices, smart phones, etc. and in stream ciphers, protocol design, etc. in software areas. Hence, the CSPRNG design is compatible to both-hardware and software applications. For software development of the cipher, a parallel computing environment has been chosen because in today's computing trends, multicore processors are superseding the sequential ones; hence the primary engine for processor performance growth is to increase parallelism rather than increasing the clock rate. The paper thus presents the CSPRNG model based on hardware and software co-simulation, using a generic approach. A mathematical model of the PRNG is designed based on above specifications and is mathematically proven to be resistant against various LFSR-based attacks.
展开▼