Paper 2025/1716

Beyond Shannon: Operational Perfect Secrecy as a Generalised Model for Information-Theoretic Security

Abstract

Shannon’s 1949 theorem defines perfect secrecy as a condition where every possible message remains equally likely given any ciphertext, which requires a key at least as long as the message. This definition, while foundational, is binary and assumes uniform message priors—assumptions rarely met in real communication systems. It cannot express the fact that secrecy degrades gradually as key entropy decreases, and it does not account for semantic structure or contextual knowledge available to adversaries. This paper extends Shannon’s framework by introducing Operational Perfect Secrecy (OPS), which defines secrecy in terms of adversarial success probability rather than requiring complete message-space coverage. Within this framework we also define two new forms of information-theoretic security: Combinatorial ITS (C-ITS), which achieves OPS through combinatorial ambiguity of candidate decryptions, and Dimensional Ambiguity ITS (DA-ITS), which achieves OPS by concealing the dimensionality of the key space itself. We show that OPS converges to Shannon secrecy when the support size approaches the message space, while providing meaningful guarantees even with shorter keys. These results generalise the concept of perfect secrecy into a continuous, operational measure and establish a new theoretical foundation for scalable information-theoretic security.