Cryptanalysis

Also known as: codebreaking

Cryptanalysis is the study of analyzing and breaking cryptosystems — recovering the plaintext, the key, or the internal structure of an algorithm without being handed the key.¹ It is the analytic counterpart to cryptography: one side builds, the other probes, and progress on each drives the other.

Cryptanalysis works from intercepted data toward the secret the cipher was meant to hide.

How it works

Cryptanalysts almost always assume Kerckhoffs’s premise — that the algorithm is known and only the key is secret — so the work is to exploit any structure the cipher leaks. Classic attacks are named by what the analyst can obtain:

Ciphertext-only — only intercepted ciphertext is available, as in classic frequency analysis of a substitution cipher.
Known-plaintext — some plaintext/ciphertext pairs are known.
Chosen-plaintext — the analyst can encrypt inputs of their choosing.
Brute force — exhaustively trying keys, bounded by the key space.

Deeper techniques attack a cipher’s mathematics directly — differential and linear analysis, algebraic modeling, and solver-driven approaches that treat recovery as a constraint problem. A scheme that relies on the method staying secret is not really analyzed but merely reverse-engineered; that is obfuscation, not encryption.

Variants

The strength of an attack is judged by two things: the resources it needs (data, chosen queries, memory) and the work factor — how much faster than brute force it runs. A “break” in the academic sense may still be wholly impractical (e.g. requiring 2¹²⁶ operations against a 128-bit key), yet it signals structural weakness. Attacks also target the implementation rather than the algorithm: side-channel analysis reads timing, power, or electromagnetic emissions, and protocol-level flaws (key reuse, predictable IVs, weak random number generators) are where real systems most often fall.

In practice

Historically, cryptanalysis decided wars — the breaking of the Enigma and Lorenz ciphers is the canonical example, and it is where systematic codebreaking, statistical attacks, and early computing converged. Today the same discipline vets civilian standards: a cipher like AES earns trust precisely because decades of public cryptanalysis have failed to dent it, while weaker or aging designs (DES, RC4) are retired as attacks accumulate.

Relevance to SDR

Cryptanalysis in the radio world is mostly about understanding formats, not breaking strong ciphers. Recovering an undocumented on-air transformation — framing, scrambling, or an obfuscation layer — is a cryptanalytic exercise carried out entirely from public observation. GopherTrunk’s clean-room work on the Motorola P25 talker-alias scheme (issue #773) is a worked example: candidate models such as shift-register and round-function update rules were hypothesized and tested against captured data until the actual substitution table emerged, all without reference to any third-party source code. Properly encrypted voice — P25 AES-256, DMR RC4, or TETRA TEA — is not in scope: the math is sound and the key is absent.

Sources

Cryptanalysis — Wikipedia, for attack models and the assumption that the algorithm is public. ↩