Leakage-Resilient Incompressible Cryptography: Constructions and Barriers

What it is

A theory paper (ePrint 2026/2461, a minor revision of the ASIACRYPT 2024 paper) defining a security model that combines two side-channel models. In leakage resilience, the adversary sees part of the key plus the whole ciphertext or signature. In incompressibility, it sees the whole key but only part of the ciphertext or signature. The abstract says a scheme secure in one model can fail in the other. No implementation or benchmarks are given.

Parameters the summary omits

• Incompressible SKE (information-theoretic): ciphertext-rate 1/2 and compression-rate 1/2. The abstract presents this as an improvement over Dziembowski (CRYPTO 2006), the existing information-theoretic baseline, and says it improves the final parameters of the SKE-to-leakage-resilient transformation and the PKE construction.
• Signature (classical random oracle model, assuming DCR or LWE, plus Decisional Linear): signature-rate and compression-rate 1-o(1), and leakage-rate 1.
• PKE: built from an incompressible SKE plus a new leakage-resilient non-committing KEM (LR-NC-KEM). The abstract gives no PKE rates.

Preconditions / Caveats

• The signature scheme depends on the random oracle heuristic.
• The impossibility results rule out basing improved schemes on black-box reductions. They do not rule out non-black-box constructions.
• The abstract names LR-NC-KEM and invertible extractors as possibly reusable tools, but states no applications.

What to do

Nothing to deploy. If you are evaluating leakage-resilient or incompressible designs, use the rates above as the current reference points, and read the full paper for the concrete constructions.