Authentication codes in plaintext and chosen-content attacks

Abstract

We study authentication codes (A-codes) where it is assumed that the enemy has access to the content of the intercepted cryptogram. This is similar to plaintext attack in secrecy systems. Enemy's success is defined in two ways. The first is as in Simmons' model. We will also consider chosen- content attacks in which the success is by constructing a fraudulent cryptogram with a given content. We will obtain information theoretic bounds, define perfect protection and obtain lower bounds on the number of encoding rules for codes with perfect protection against chosen-content impersonation and chosen-content plaintext substitution. We characterize these A-codes when the number of encoding rules is minimum. We give methods for making an' A-code resistant against plaintext and chosen-context plaintext attack.