We present XFormer, a novel human mesh and motion capture method that achieves real-time performance on consumer CPUs given only monocular images as input. The proposed network architecture contains two branches: a keypoint branch that estimates 3D human mesh vertices given 2D keypoints, and an image branch that makes prediction directly from the RGB image features. At the core of our method is a cross-modal transformer block that allows information flow across these two branches by modeling the attention between 2D keypoint coordinates and image spatial features. Our architecture is smartly designed, which enables us to train on various types of datasets including images with 2D/3D annotations, images with 3D pseudo labels, and motion capture datasets that do not have associated images. This effectively improves the accuracy and generalization ability of our system. Built on a lightweight backbone (MobileNetV3), our method runs blazing fast (over 30fps on a single CPU core) and still yields competitive accuracy. Furthermore, with a HRNet backbone, XFormer delivers state-of-the-art performance on Huamn3.6 and 3DPW datasets.

Abstract Preparing a large number of physical traces is an important first step in Side-Channel Analysis, especially in Deep-Learning based Side-Channel Analysis (DL-SCA). With sufficient training data and a proper modeling algorithm, the secret key of cryptographic devices can be successfully recovered with a small number of attacking data. However, in reality, it may be impossible or difficult, in some threat models, to collect sufficient data due to various resource constraints. In this case, the performance of DL-SCA will be severely decreased. In this work, we propose an easy-to-implement method to achieve an efficient DL-SCA with a small number of training data in the scenario of software-based cryptographic implementations. Our simultaneously multi-byte training method, which trains the model with side-channel leakage characteristics of different byte intermediate values, significantly enhances the robustness and performance of DL-SCA. The simulated experiment shows that our method achieves more robust profiling. The success rate of recovering a secret AES key can be improved by 250% with the same collected data. The results of attacking real-world COTS USIM cards are consistent with the ones of simulation-based counterparts. Compared with state-of-the-art data-augmentation techniques, our results show that the proposed method can achieve the same or even better performance without additional generated training data.

Abstract Side-channel analysis (SCA) has become an increasing important method to assess the physical security of cryptographic systems. In the process of SCA, the number of attack data directly determines the performance of SCA. With sufficient attack data, the adversary can achieve a successful SCA. However, in reality, the cryptographic device may be protected with some countermeasures to limit the number of encryptions using the same key. In this case, the adversary cannot use casual numbers of data to perform SCA. The performance of SCA will be severely dropped if the attack traces are insufficient. In this paper, we introduce wavelet scatter transform (WST) and short-time fourier transform (STFT) to non-profiled side-channel analysis domains, to improve the performance of side-channel attacks in the context of insufficient data. We design a practical framework to provide suitable parameters for WST/STFT-based SCA. Using the proposed method, the WST/STFT-based SCA method can significantly enhance the performance and robustness of non-profiled SCA. The practical attacks against four public datasets show that the proposed method is able to achieve more robust performance. Compared with the original correlation power analysis (CPA), the number of attack data can be reduced by 50–95%.

Side-channel security has become a significant concern in the NIST post-quantum cryptography standardization process. The lattice-based CRYSTALS-Dilithium (abbr. Dilithium) becomes the primary signature standard algorithm recommended by NIST for most use cases in July 2022 due to its excellent performance in security and efficiency. Compared to Dilithium’s rich theoretical security analysis results, the side-channel security of its physical implementations needs to be further explored. In 2021, Liu et al. proposed a two-stage randomness leakage attack against Dilithium, in which only one randomness bit with a probability$> 0.5$per signature is enough to recover the private key. However, they only carried out proof-of-concept experiments on “research-oriented” reference implementation of polynomial addition. Whether this method applies to complete real-world implementations of Dilithium is unknown. In this paper, we put this randomness leakage attack into real-world and recover the private key of unprotected and masked Dilithium on Arm Cortex-M4 processor using non-profiled power analysis attacks. Since randomness is introduced in the signing process, it is challenging to recover the randomness bit of Dilithium with high success rate in only one trace. Inspired by Liu et al., we propose a new non-profiled attack called Public Template Attack (PTA), a template-attack-like method that builds templates using public information. With PTA, we recover the randomness bit of unprotected and masked Dilithium with a success rate of 95% and 62% in one power trace, respectively. To demonstrate practicality, we perform practical power analysis attacks against different security levels of round 3 unprotected and masked Dilithium on STM32F405 microprocessor. Using 10,000 traces, the private key of unprotected Dilithium2 is recovered in 0.5 hours with an ordinary PC desktop. Our attack is 240 times faster than the state-of-the-art non-profiled attack. Moreover, the private key of masked Dilithium2 is recovered using 680,000 traces in 38 hours. To the best of our knowledge, we are the first to successfully attack masked Dilithium using non-profiled attacks.

Abstract Side-channel resistance is nowadays widely accepted as a crucial factor in deciding the security assurance level of cryptographic implementations. In most cases, non-linear components (e.g. S-Boxes) of cryptographic algorithms will be chosen as primary targets of side-channel attacks (SCAs). In order to measure side-channel resistance of S-Boxes, three theoretical metrics are proposed and they are reVisited transparency order (VTO), confusion coefficients variance (CCV), and minimum confusion coefficient (MCC), respectively. However, the practical effectiveness of these metrics remains still unclear. Taking the 4-bit and 8-bit S-Boxes used in NIST Lightweight Cryptography candidates as concrete examples, this paper takes a comprehensive study of the applicability of these metrics. First of all, we empirically investigate the relations among three metrics for targeted S-boxes, and find that CCV is almost linearly correlated with VTO, while MCC is inconsistent with the other two. Furthermore, in order to verify which metric is more effective in which scenarios, we perform simulated and practical experiments on nine 4-bit S-Boxes under the non-profiled attacks and profiled attacks, respectively. The experiments show that for quantifying side-channel resistance of S-Boxes under non-profiled attacks, VTO and CCV are more reliable while MCC fails. We also obtain an interesting observation that none of these three metrics is suitable for measuring the resistance of S-Boxes against profiled SCAs. Finally, we try to verify whether these metrics can be applied to compare the resistance of S-Boxes with different sizes. Unfortunately, all of them are invalid in this scenario.

Abstract We revisit the definition of transparency order (TO) and that of modified transparency order (MTO) as well, which were proposed to measure the resistance of substitution boxes (S-boxes) against differential power analysis (DPA). We spot a definitional flaw in original TO, which is proved to significantly affect the soundness of TO. Regretfully, MTO overlooks this flaw, yet it happens to incur no bad effects on the correctness of MTO, even though the start point of this formulation is highly questionable. It is also this neglect that made MTO consider a variant of multi-bit DPA attack, which was mistakenly thought to appropriately serve as an alternative powerful attack. This implies the soundness of MTO is also more or less arguable. Therefore, we fix this definitional flaw and provide a revised definition named reVisited TO (VTO). For demonstrating validity and soundness of VTO, we present simulated and practical DPA attacks on implementations of $4\times 4$ and $8\times 8$ S-boxes. In addition, we also illustrate the soundness of VTO in masked S-boxes. Furthermore, as a concrete application of VTO, we present the distribution of VTO values of optimal affine equivalence classes of $4\times 4$ S-boxes and give some recommended guidelines on how to select $4\times 4$ S-boxes with higher DPA resistance at the identical level of implementation cost.