Dangerous Functions for Feedforward Pseudo-Noise Generators with Memory
Andrea Paiusco · Journal of Information and Optimization Sciences · 1991
Purely feedforward pseudo-noise generators with memory can be used to build stream ciphers which are immune to correlation attacks. Nevertheless, an unwise choice of the security amplifier’s function may allow a known-plaintext attack that produces an approximate solution; this attack is much faster than the search over most of the key space which is otherwise necessary. In the following we put forward a design criterium which makes the attack unfeasible.