Computationally secure information flow

This thesis presents a definition and a static program analysis for secure information flow. The definition of secure information flow is not based on non-interference, but on the computational independence of the program’s public outputs from its secret inputs. Such definition allows cryptographic primitives to be gracefully handled, as their security is usually defined to be only computational, not information-theoretical. The analysis works on a simple imperative programming language containing a cryptographic primitive—encryption—as a possible operation. The analysis captures the intuitive qualities of the (lack of) information flow from a plaintext to its corresponding ciphertext. We prove the analysis correct with respect to the definition of secure information flow described above. In the proof of correctness we assume that the encryption primitive hides the identity of plaintexts and keys. This thesis also considers the case where the identities of plaintexts and keys are not hidden by encryption, i.e. given two ciphertexts it may be possible to determine whether the corresponding plaintexts are equal or not. We also give an analysis for this case, though it is not a whole program analysis. Namely, we cannot analyse loops. Nevertheless, with the help of the analysis one can check, whether two formal expressions (which are equivalent to the output of programs without loops) have indistinguishable interpretations as bit-strings.