The padding scheme for RSA signatures

The RSA scheme is used to sign messages; however, in order to avoid forgeries, a message can be padded with a fixed string of data P. De Jonge and Chaum showed in 1985 that forgeries can be constructed if the size of P (measured in bytes) is less than the size of N/3, where N is the RSA modulus. Girault and Misarsky then showed in 1997 that forgeries can be constructed if the size of P is less than the size of N/2. In 2001, Brier, Clavier, Coron and Naccache showed that forgeries can still be constructed when the size of P is less than two thirds the size of N. In this paper, we demonstrate that this padding scheme is always insecure; however, the complexity of actually finding a forgery is O(N). We then focus specifically on the next unsettled case, where P is less than 3/4 the size of N and show that finding a forgery is equivalent to solving a set of diophantine equations. While we are not able to solve these equations, this work may lead to a break-through by means of algebraic number theory techniques.