Lock-in effect in cascades of clock-controlled shift-registers

Abstract

Cascaded cryptographic keystream generators as proposed by Gollmann possess a cryptanalytic weakness termed “lock-in” in this article. If the initial state has been guessed correctly apart from its phase a decryption cascade can be set up in which the effects of each stage of the original cascade are unravelled in reverse order. Once the decryption cascade has “locked in” on the original cascade, the state of the latter is known, and hence its future output and its output in the remote past. This weakness is studied; its effects are readily mitigated by taking certain precautions. Lock-in may also be used constructively as a synchronization technique.