We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Cryptography and Security

Title: LRPC codes with multiple syndromes: near ideal-size KEMs without ideals

Abstract: We introduce a new rank-based key encapsulation mechanism (KEM) with public key and ciphertext sizes around 3.5 Kbytes each, for 128 bits of security, without using ideal structures. Such structures allow to compress objects, but give reductions to specific problems whose security is potentially weaker than for unstructured problems. To the best of our knowledge, our scheme improves in size all the existing unstructured post-quantum lattice or code-based algorithms such as FrodoKEM or Classic McEliece. Our technique, whose efficiency relies on properties of rank metric, is to build upon existing Low Rank Parity Check (LRPC) code-based KEMs and to send multiple syndromes in one ciphertext, allowing to reduce the parameters and still obtain an acceptable decoding failure rate. Our system relies on the hardness of the Rank Support Learning problem, a well-known variant of the Rank Syndrome Decoding problem. The gain on parameters is enough to significantly close the gap between ideal and non-ideal constructions. It enables to choose an error weight close to the rank Gilbert-Varshamov bound, which is a relatively harder zone for algebraic attacks. We also give a version of our KEM that keeps an ideal structure and permits to roughly divide the bandwidth by two compared to previous versions of LRPC KEMs submitted to the NIST with a Decoding Failure Rate (DFR) of $2^{-128}$.
Comments: 31 pages
Subjects: Cryptography and Security (cs.CR)
Cite as: arXiv:2206.11961 [cs.CR]
  (or arXiv:2206.11961v1 [cs.CR] for this version)

Submission history

From: Victor Dyseryn [view email]
[v1] Thu, 23 Jun 2022 20:20:30 GMT (170kb)

Link back to: arXiv, form interface, contact.