Some Information-Theoretic Problems of Discrete Data Protection

We consider the secrecy of discrete data, introduced by Shannon [Bell Syst. Tech. J., 28, No. 4, 656–715 (1949)]. We study the properties of certain key sets for a given probability distribution of messages, and methods for increasing the secrecy by source coding (in particular, by variable-to-fixed rate coding). We specify the statement of the randomization problem and show that sometimes, for instance, in the case of the uniform multiple substitution of letters, the randomization is inefficient.