The Stop-and-go-generator

Abstract

Under the correct assumptions cascading of primitive shift registers leads to interesting results. But from Gollmann’s work it is clear that general results on cascaded arbitrary shift registers cannot be expected. In order to guarantee a good statistical behaviour of the Stop-and-Go-Sequence it is suggested that the output sequence ut is finally XOR-gated with another PN-sequence. The statistical behaviour of (ut)t itself — though theoretically quite good in special cases — is so that a cryptoanalytic attackwould be promising in spite of the extremely high linear equivalent of the sequence.