First published May 15, 2024 Application Aachen patent G – Physics

METHOD AND APPARATUS FOR SECURE PROCESSING WHEN USING A HARDWARE CALCULATION UNIT

Espacenet Family 84332044

Abstract

[0001] A method and apparatus for secure processing of matrix multiplication using a hardware calculation unit (31; 63) as well as a corresponding apparatus and computer program is disclosed. The method for secure processing of a matrix multiplication using a hardware calculation unit (31; 63) adapted to perform matrix multiplications comprises receiving (S1) a matrix, encrypting (S2) the matrix by generating a permuted matrix by permuting the rows or the columns of the matrix with a first permutation vector, writing (S3) the permuted matrix to storage elements of the hardware calculation unit (31; 63) adapted to perform matrix multiplications, receiving (S4) an input vector which is to be multiplied with the matrix, and performing (S5) a matrix multiplication using the input vector and the hardware calculation unit (31; 63), wherein an output of the calculation provided by the hardware calculation unit (31; 63) is decrypted by applying an inverse of the first permutation vector to either the input vector or the output of the hardware calculation unit (31; 63).

Inventors 2

Galicia Cota, Melvin Estuardo
Galicia, Cota Melvin Estuardo

Inventors are linked to an Aachen author when surname and first name match exactly one author.

Applicants

Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen RWTH Aachen University

Cited papers 1 in the database

A Low Cost Weight Obfuscation Scheme for Security Enhancement of ReRAM Based Neural Network Accelerators
2021 Asia and South Pacific Design Automation Conference (ASP-DAC) conference-paper Computer Science Physical Unclonable Functions (PUFs) and Hardware Security

10

citations

Non-patent literature as cited (3)

- SHANSHI HUANG ET AL: "New Security Challenges on Machine Learning Inference Engine: Chip Cloning and Model Reverse Engineering", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 21 March 2020 (2020-03-21), XP081626909
- WANG YUHANG WANGYUHANG@NCEPU EDU CN ET AL: "A Low Cost Weight Obfuscation Scheme for Security Enhancement of ReRAM Based Neural Network Accelerators", PROCEEDINGS OF THE 13TH ACM SIGPLAN INTERNATIONAL SYMPOSIUM ON HASKELL, ACMPUB27, NEW YORK, NY, USA, 18 January 2021 (2021-01-18), pages 499 - 504, XP058731357, ISBN: 978-1-4503-8050-8, DOI: 10.1145/3394885.3431599
- YANG YANG ET AL: "A Comprehensive Survey on Secure Outsourced Computation and Its Applications", IEEE ACCESS, vol. 7, 28 October 2019 (2019-10-28), pages 159426 - 159465, XP011754120, [retrieved on 20191107], DOI: 10.1109/ACCESS.2019.2949782