Paper 2021/1425

Improving First-Order Threshold Implementations of SKINNY

Andrea Caforio, Daniel Collins, Ognjen Glamocanin, and Subhadeep Banik


Threshold Implementations have become a popular generic technique to construct circuits resilient against power analysis attacks. In this paper, we look to devise efficient threshold circuits for the lightweight block cipher family SKINNY. The only threshold circuits for this family are those proposed by its designers who decomposed the 8-bit S-box into four quadratic S-boxes, and constructed a 3-share byte-serial threshold circuit that executes the substitution layer over four cycles. In particular, we revisit the algebraic structure of the S-box and prove that it is possible to decompose it into (a) three quadratic S-boxes and (b) two cubic S-boxes. Such decompositions allow us to construct threshold circuits that require three shares and executes each round function in three cycles instead of four, and similarly circuits that use four shares requiring two cycles per round. Our constructions significantly reduce latency and energy consumption per encryption operation. Notably, to validate our designs, we synthesize our circuits on standard CMOS cell libraries to evaluate performance, and we conduct leakage detection via statistical tests on power traces on FPGA platforms to assess security.

Available format(s)
Publication info
Published elsewhere. INDOCRYPT-2021
DPAMaskingSKINNYThreshold Implementation
Contact author(s)
andrea caforio @ epfl ch
daniel collins @ epfl ch
ognjen glamocanin @ epfl ch
subhadeep banik @ epfl ch
2021-10-24: received
Short URL
Creative Commons Attribution


      author = {Andrea Caforio and Daniel Collins and Ognjen Glamocanin and Subhadeep Banik},
      title = {Improving First-Order Threshold Implementations of SKINNY},
      howpublished = {Cryptology ePrint Archive, Paper 2021/1425},
      year = {2021},
      note = {\url{}},
      url = {}
Note: In order to protect the privacy of readers, does not use cookies or embedded third party content.