This article, for the first time, demonstrates Cross-device Deep Learning Side-Channel Attack (X-DeepSCA), achieving an accuracy of > 99.9%, even in presence significantly higher inter-device variations compared to inter-key variations. Augmenting traces captured from multiple devices training and with proper choice hyper-parameters, proposed 256-class Neural Network (DNN) learns accurately power side-channel leakage AES-128 target encryption engine, N-trace (N ≤ 10) X-DeepSCA attack breaks...

Power side-channel analysis (SCA) has been of immense interest to most embedded designers evaluate the physical security system. This work presents profiling-based cross-device power SCA attacks using deep learning techniques on 8-bit AVR microcontroller devices running AES-128. Firstly, we show practical issues that arise in these due significant device-to-device variations. Secondly, utilizing Principal Component Analysis (PCA) based pre-processing and multi-device training, a Multi-Layer...

Mathematically secure cryptographic algorithms, when implemented on a physical substrate, leak critical "side-channel" information, leading to power and electromagnetic (EM) analysis attacks. Circuit-level protections involve switched capacitor, buck converter, or series low-dropout (LDO) regulator-based implementations, each of which suffers from significant power, area, performance tradeoffs has only achieved minimum traces disclosure (MTD) 10M till date. Utilizing an in-depth white-box...

Computationally-secure cryptographic algorithms when implemented on physical platforms leak critical signals correlated with the secret key in form of power consumption and electromagnetic (EM) emanations. This can be exploited by an adversary, leading to side-channel attacks (SCA) that recover key. Circuit-level on-chip countermeasures include a switched-capacitor current equalizer [1], charge-recovery logic [2], integrated voltage regulator (IVR) [3], all-digital low-dropout (LDO) [4],...

This work presents a Cross-device Deep-Learning based Electromagnetic (EM-X-DL) side-channel analysis (SCA) on AES-128, in the presence of significantly lower signal-to-noise ratio (SNR) compared to previous works. Using novel algorithm intelligently select multiple training devices and proper choice hyperparameters, proposed 256-class deep neural network (DNN) can be trained efficiently utilizing pre-processing techniques like PCA, LDA, FFT measurements from target encryption engine running...

This letter describes a Delta-Sigma ADC based Power-Side-Channel-Attack Sensor. Use of 64 sampling capacitors allows the use over-sampling architecture even with decoupling capacitor connected to power supply. The LDO low-leakage S/H is used as driver for integrator's amplifier minimize offset error. A differential conversion method utilizing Dual-Integrate (CAPs) provides signal processing compensate drift due supply voltage (VDD) variations. prototype sensor chip fabricated in 65nm CMOS...

As machine learning (ML) permeates fields like healthcare, facial recognition, and blockchain, the need to protect sensitive data intensifies. Fully Homomorphic Encryption (FHE) allows inference on encrypted data, preserving privacy of both ML model. However, it slows down non-secure by up five magnitudes, with a root cause replacing non-polynomial operators (ReLU MaxPooling) high-degree Polynomial Approximated Function (PAF). We propose SmartPAF, framework replace low-degree PAF then...

This article, for the first time, demonstrates an efficient circuit-level countermeasure to prevent deep-learning based side-channel analysis (DLSCA) attacks on encryption devices. Machine learning (ML) SCA, particularly DLSCA have been shown be extremely effective as it can potentially reveal secret key of cryptographic device with low a single trace, by offloading heavy-lifting profiling phase where model learns correlated leakage patterns key. work presents current-domain signature...

This paper presents SNOW-SCA, the first power side-channel analysis (SCA) attack of a 5G mobile communication security standard candidate, SNOW-V, running on 32-bit ARM Cortex-M4 microcontroller. First, we perform generic known-key correlation (KKC) to identify leakage points. Next, (CPA) is performed, which reduces complexity two key guesses for each byte. The correct secret then uniquely identified utilizing linear discriminant (LDA). profiled SCA with LDA achieves 100% accuracy after...

Countermeasures against power side-channel attack (SCA) range from algorithmic modifications to modifying the consumption characteristics of device by inclusion noise sources or suppressing current signature. However, these techniques are, in most cases, passive protection and come with unacceptable area overheads for resource-constrained devices. In this letter, we propose a SCA detection sensor based on 1-b delta-sigma analog-to-digital converter, which allows an ongoing within as fast...

In this work, we present a comprehensive analysis of explainability Neural Network (NN) models in the context power Side-Channel Analysis (SCA), to gain insight into which features or Points Interest (PoI) contribute most classification decision. Although many existing works claim state-of-the-art accuracy recovering secret key from cryptographic implementations, it remains be seen whether actually learn representations leakage points. evaluated reasoning behind success NN model, by...

Manufacturing variability demonstrates significant variations in dynamic power consumption profiles during program execution, even if the printed circuit boards (PCB) are identical and processors execute same operations on data. In this work, we show how can be leveraged to benefit of manufacturers by utilizing machine learning (ML) based PCB identification. The proposed technique achieves 100% accuracy identifying PCBs from their traces after training a linear discriminant analysis (LDA)...

This work presents a power side-channel analysis (SCA) of lightweight cryptography (LWC) algorithm, XOODYAK, implemented on an FPGA. First, we perform generic leakage detection tests for two phases authenticated encryption with associated data (AEAD) mode, namely INITIALIZE, and ABSORB. Second, develop novel hypothetical attack models correlation (CPA) demonstrate success rate (SR) 92%/82% minimum-traces-to-disclosure (MTD)=13K/38K the INITIALIZE/ABSORB phases, respectively. Third, evaluate...

Abstract Machine learning (ML) is getting more pervasive. Wide adoption of ML in healthcare, facial recognition, and blockchain involves private sensitive data. One the most promising candidates for inference on encrypted data, termed Fully Homomorphic Encryp-tion (FHE), preserves privacy both data model. However, it slows down plaintext by six magnitudes, with a root cause replacing non-polynomial operators latency-prohibitive 27-degree Polynomial Approximated Function (PAF). While prior...

This paper presents SNOW-SCA, the first power side-channel analysis (SCA) attack of a 5G mobile communication security standard candidate, SNOW-V, running on 32-bit ARM Cortex-M4 microcontroller. First, we perform generic known-key correlation (KKC) to identify leakage points. Next, (CPA) is performed, which reduces complexity two key guesses for each byte. The correct secret then uniquely identified utilizing linear discriminant (LDA). profiled SCA with LDA achieves 100% accuracy after...

In this work, we present a high-performance architecture for Discrete Gaussian (DG) sampling used in lattice-based cryptography (LBC) through lookup table (LUT) optimization the combinational datapath as well FPGA aware pipelining to reduce resource utilization and decrease latency. The proposed tree-splitting technique translates discrete distribution generating (DDG) tree Knuth-Yao-based non-uniform into LUT-based logic with pipelined architecture. This allows reduction area-time product...

This work presents a Cross-device Deep-Learning based Electromagnetic (EM-X-DL) side-channel analysis (SCA), achieving >90% single-trace attack accuracy on AES-128, even in the presence of significantly lower signal-to-noise ratio (SNR), compared to previous works. With an intelligent selection multiple training devices and proper choice hyperparameters, proposed 256-class deep neural network (DNN) can be trained efficiently utilizing pre-processing techniques like PCA, LDA, FFT target...

Traditional public-key cryptographic schemes are soon going to be replaced with Post-Quantum Cryptographic (PQC) ensure security guarantees in a Quantum Computing-enabled world. While Computing will help solve many hard problems intractable by classical computing paradigm, it also compromise the that traditional built upon. Among National Institute of Standards and Technology (NIST) finalist PQC schemes, SABER Key Encapsulation Mechanism (KEM) is only one based on Module Learning With...

In order to improve the reliability of SRAM-based field programmable gate array (FPGA) in various harsh environments, an FPGA online fault location method with high accuracy needs studying support subsequent tolerance technologies.To achieve location, it is necessary overcome a series limitations such as lack hardware resources and user design knowledge.Therefore, this paper proposes black box based on meshed bitstream copy.Generally, linear feedback shift register (LFSR) used...