20.500.12556/RUP-23063 PN-SCA a high generalization and fast profiled SCA based on prototypical networks Recently, there has been a growing interest in scenarios that employ a few (imbalanced) power traces for deep learning based side-channel analysis (DL-SCA). Finding a lightweight DL-SCA architecture that is more generalizable and faster to learn is of great importance when handling such situations. In this work, an initial analysis is conducted on the capability of learning and the susceptibility to attacks of prevalent DL-SCA models, focusing on cases that culminate in an unsuccessful attack. Subsequently, a meta-learning technique, known as prototypical networks, is delineated for the construction of a lightweight deep learning framework. In contrast to conventional DL-SCA models, the proposed architecture, designated as PN-SCA, does not predict the probability of a power trace belonging to an intermediate value in classification. Instead, it facilitates the learning of an encoder capable of embedding power consumption data within a latent space, while also establishing templates, or prototypes, for diverse categories. Moreover, we developed a metric that is specifically intended for the selection of hyperparameters due to the unique training phase of PN-SCA. Finally, four distinct scenarios are established with few (including one imbalanced) power traces to evaluate the correctness of our architecture. The results clearly illustrate the advantages of our PN-SCA in terms of generalization, reduced training costs (with a decrease in profiling traces by over 50%), and significantly enhanced attack effect (with a reduction of power traces requirements by over 90%), thus demonstrating notable improvements over existing methodologies V zadnjem času narašča zanimanje za scenarije, ki pri globokem učenju za analizo stranskih kanalov (DL-SCA) uporabljajo majhno število (neuravnoteženih) energetskih sledi. Iskanje lahke arhitekture DL-SCA, ki je bolj posplošljiva in se hitreje uči, je pri obravnavi takšnih situacij zelo pomembno. V tem delu je najprej izvedena analiza učne sposobnosti in dovzetnosti za napade pri razširjenih modelih DL-SCA, s poudarkom na primerih, ki se končajo z neuspešnim napadom. Nato je predstavljena tehnika meta-učenja, znana kot prototipske mreže, za izdelavo lahkega ogrodja globokega učenja. V nasprotju s konvencionalnimi modeli DL-SCA predlagana arhitektura, poimenovana PN-SCA, ne napoveduje verjetnosti, da energetska sled pripada določeni vmesni vrednosti pri klasifikaciji. Namesto tega omogoča učenje kodirnika, ki lahko podatke o porabi energije preslika v latentni prostor, hkrati pa vzpostavi predloge oziroma prototipe za različne kategorije. Poleg tega smo razvili metriko, posebej namenjeno izbiri hiperparametrov zaradi edinstvene faze učenja PN-SCA. Nazadnje so bili vzpostavljeni štirje različni scenariji z majhnim številom energetskih sledi (vključno z enim neuravnoteženim), da bi ocenili pravilnost naše arhitekture. Rezultati jasno prikazujejo prednosti našega pristopa PN-SCA v smislu boljše posplošitve, zmanjšanih stroškov učenja (z zmanjšanjem profilnih sledi za več kot 50 %) ter bistveno izboljšanega učinka napada (z zmanjšanjem potrebnega števila energetskih sledi za več kot 90 %), kar kaže na opazne izboljšave v primerjavi z obstoječimi metodami. side-channel analysis deep learning based SCA prototypical networks few-shot learning analiza stranskih kanalov analiza stranskih kanalov na osnovi globokega učenja prototipske mreže true false true Angleški jezik Slovenski jezik Članek v reviji 2026-05-20 11:33:24 2026-05-20 11:33:28 2026-05-21 03:10:11 0000-00-00 00:00:00 2026 0 0 str. 77-94 iss. 1 Vol. 42 Feb. 2026 0000-00-00 PostprintKoncna NiDoloceno NiDoloceno 0000-00-00 0000-00-00 2026-02-12 004.8 1573-0727 10.1007/s10836-026-06219-4 278763779 RAZ_Ou_Yu_2026.pdf RAZ_Ou_Yu_2026.pdf 1 CFC7C8B2CE35B2FA90F448098D651F9F 5d3ae2f29254b33720807a99f1a37ec64cb8ba33269d9f2bf9b8adab63855612 0b38f1dd-542f-11f1-9e8d-005056ac49c0 https://repozitorij.upr.si/Dokument.php?lang=slv&id=33703 https://link.springer.com/article/10.1007/s10836-026-06219-4 1 db610c3d-542e-11f1-9e8d-005056ac49c0 https://repozitorij.upr.si/Dokument.php?lang=slv&id=33702 Inštitut Andrej Marušič 0 0 0