Unicity distance is the ciphertext-length scale at which a specified cipher and plaintext model are expected to leave essentially one plausible cryptographic key. Claude Shannon obtained an approximate threshold from the sharp decrease of key equivocation in a random-cipher model; it is not a universal length guaranteeing zero residual conditional entropy for every ciphertext.[ref]
Communication Theory of Secrecy Systems by Claude Shannon. Sections 14–15, printed pp. 691–695, especially the discussion after Figure 7
For a realized ciphertext, uniqueness depends on which candidate decryptions the plaintext model allows. A probability distribution giving every finite plaintext positive mass need not exclude every wrong cryptographic key at any finite length. A source symmetry can even preserve ambiguity forever. The source redundancy in cryptanalysis estimate therefore needs its source and cipher assumptions.
Unicity distance is a concept in cryptography that refers to the minimum amount of ciphertext required to ensure that a given ciphertext corresponds to exactly one possible plaintext. In other words, it is the length of ciphertext needed to guarantee that there is a unique plaintext that could produce that ciphertext using a particular encryption scheme. In contexts like symmetric encryption, the unicity distance is important for assessing the security of a cryptosystem.
New to topics? Read the docs here!