Primjena strojnog učenja u napadima na sporedne kanale

master's thesis

• Title (english): Application of machine learning in side channel attacks
• Author: Magdalena Mucić
• Mentor: Marko Čupić (mentor)
• Committee member: Marko Čupić (predsjednik povjerenstva)
• Committee member: Stjepan Groš (član povjerenstva)
• Committee member: Siniša Šegvić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-12, Croatia

Abstract

Područje kriptoanalize ima važnu ulogu u kriptografiji na kojoj se temelji sigurnost svih modernih algoritama kriptiranja: ispituje i proučava kriptografske algoritme te pokušava pronaći njihove ranjivosti i mane. Jedan od pristupa u kriptoanalizi su i napadi na sporedne kanale kojima je cilj pronaći ranjivosti koje proizlaze iz sklopovske implementacije algoritma. Postoji više pristupa izvođenja napada na sporedne kanale, a neki od njih potpomažu se i strojnim učenjem. U okviru ovoga rada istražena je primjena strojnog učenja u kriptografiji, posebno se fokusirajući na dva široko korištena algoritma kriptiranja kao predstavnika dviju velikih grana kriptografije: EdDSA, kao predstavnik asimetrične kriptografije, i AES, kao predstavnik simetrične kriptografije. Pokazano je da modeli strojnog učenja mogu postići visoku točnost i entropiju pogađanja kod predviđanja točnog ključa koristeći mjerenja prikupljena napadima na sporedne kanale.

Abstract (english)

The field of cryptanalysis plays an important role in cryptography, on which the security of all modern encryption algorithms is based: it examines and studies cryptographic algorithms and tries to find their vulnerabilities and flaws. One of the approaches in cryptanalysis are side channels attacks, the aim of which is to find vulnerabilities resulting from the implementation of the algorithm. There are multiple approaches to performing side-channel attacks, some of which use machine learning. This thesis explores the application of machine learning in cryptography, focusing especially on two widely used encryption algorithms as representatives of two major branches of cryptography: EdDSA, as a representative of asymmetric cryptography, and AES, as a representative of symmetric cryptography. Experiments show that machine learning models can achieve high accuracy and guessing entropy in predicting the correct key using traces collected from side-channel attacks.

• Keywords: Strojno učenje; napadi na sporedne kanale; AES; EdDSA; konvolucijske neuronske mreže
• Keywords (english): Machine learning; side channel attacks; AES; EdDSA; Convolutional neural networks
• Language: croatian
• URN:NBN: urn:nbn:hr:168:396933
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-10-11 09:54:00