logo SBA

ETD

Archivio digitale delle tesi discusse presso l’Università di Pisa

Tesi etd-09092026-172232


Tipo di tesi
Tesi di laurea magistrale
Autore
STEFANI, ANDREA
URN
etd-09092026-172232
Titolo
FPGA Design of a Loosely-Coupled Accelerator for the CROSS Post-Quantum Digital Signature Scheme
Dipartimento
INGEGNERIA DELL'INFORMAZIONE
Corso di studi
INGEGNERIA ELETTRONICA
Relatori
.
relatore Fanucci, Luca
relatore Nannipieri, Pietro
Parole chiave
  • cross
  • digital signature
  • fpga
  • hardware accelerator
  • post-quantum cryptography
  • pqc
Data inizio appello
01/10/2026
Consultabilità
Non consultabile
Data di rilascio
01/10/2096
Riassunto (Inglese)
The prospect of large-scale quantum computers threatens conventional public-key cryptography. In response, the National Institute of Standards and Technology (NIST) launched a standardization process for additional post-quantum digital signature schemes, in which CROSS was evaluated as a code-based candidate relying on the Restricted Syndrome Decoding problem. This thesis presents a loosely coupled FPGA accelerator for the CROSS v2.2 signing procedure. The accelerator supports the R-SDP and R-SDP(G) variants and all three NIST security categories in the fast optimization corner. The parameterized RTL architecture consists of an orchestrator, shared on-chip memory, and three specialized engines for finite-field arithmetic, SHAKE-based hashing and cryptographically secure pseudorandom generation (CSPRNG), and data packing and alignment. The complete signing flow was validated against official CROSS known-answer tests. Depending on the variant and security category, post-route results on an AMD Zynq UltraScale+ FPGA show 24.6 to 25.9 thousand LUTs, 6.8 to 7.4 thousand flip-flops, maximum clock frequencies between 93.8 and 122.0 MHz, and signature generation times of 1.22-8.23 ms. Compared with the state of the art, the design offers a suitable solution for resource-constrained targets, with lower logic utilization at the cost of higher signing latency. Its parameterization allows some adjustment of the trade-off between resource utilization and performance.
Riassunto (Italiano)
File