Probabilistic Cycle Detection for Schneier's Solitaire Keystream Algorithm

Abstract

Pencil-and-paper ciphers are plausible solutions that could provide lightweight protection to the communication of resource-constrained devices. A good example in this category is Schneier's Solitaire cipher. In this paper, we propose a probabilistic solution that is able to estimate Solitaire's keystream cycle length. We also present a variation of Solitaire's original design, and evaluate the resulting construction in terms of predictability. We conduct statistical randomness tests on both the original design and the modified version based on the NIST randomness test suite. The results show that our approach improves the randomness of original Solitaire's output sequences.