Nonrandomization: a trial and error cryptanalysis directive

Generators with nonlinear functions have been proposed for cryptographic applications as running key generators in stream ciphers. It is suspected that these types of generators suffer from an undesirable nonrandomization feature, where, their output sequences contain a high rate of redundant constructs. In this paper it is shown that the possibility of breaking these ciphers can be assessed. Here, we present a method for evaluating the solving power using the sequences distribution. The given method has shown significant reduction to the cost of evaluating the solving power values. Also, it is found that as the degree of randomization increases, the solving power will decrease and vice-versa. Hence, we conclude that cryptographic secure pseudo-random sequence generators (CSPRSGs) should be designed in a manner that leads to production of sequences satisfying high randomization. This leads to constraints for the nonlinear combining functions that are used by the candidate generators.