Cryptographic devices are sensitive to side-channel attacks, which inevitably leak electromagnetic radiation, power consumption, time, and other physical information during execution. The side-channel storage vulnerability caused by storage instructions has become one of the main targets for attackers, posing a serious threat to the implementation security of cryptographic algorithms. In this article, following reveals the essence of the side-channel storage vulnerability at the computer architecture level, two novel technologies, by reducing the correlation between the processed data and emissions, are proposed to defend such attacks, namely the randomizing Hamming weight scheme and the balancing Hamming weight scheme. Furthermore, we apply the proposed scheme to AES and CRYSTALS-Kyber on the Cortex-M4 CPU. The experimental results show that this strategy can effectively eliminate the side-channel storage vulnerability at a low cost of time and space, thereby ensuring the secure implementation of cryptographic algorithms.

With the rapid development of quantum computing technology, the practical deployment of post-quantum cryptographic algorithms faces severe hardware security challenges. As a selected algorithm in the NIST post-quantum cryptography standardization project, the code-based Hamming Quasi-Cyclic (HQC) key encapsulation mechanism has been mathematically proven to possess quantum-resistance. However, its physical implementation could be vulnerable to side-channel attacks. This paper first reveals the side-channel vulnerability of the critical operation v-uy in the HQC decapsulation process. By precisely constructing ciphertexts and exploiting the Hamming weight leakage of the vect_add operation’s computational results, efficient recovery of the HQC-128 secret key is achieved. Experimental results show that with the highest optimization level of the compiler (-O3), the full key extraction can be completed with only 3,500 power traces. This finding not only provides positive analysis for the side-channel security of HQC but also offers a crucial warning for the physical layer security design of post-quantum cryptographic algorithms. The research findings highlight the necessity of ’Algorithm-Implementation-Physics’ collaborative security assessment in the post-quantum cryptography standardization process.

Traditional deep learning models have shown remarkable performance in side-channel analysis, yet their inherent black-box nature limits the understanding of their internal decision-making processes. In this paper, we propose SG-SCA, an interpretable deep learning side-channel analysis model. The model introduces a novel ANOVA F-statistic-based differential evaluation method to select shapelets with high representational power, enabling high-quality data transformation that effectively replaces the original side-channel data. Furthermore, by integrating the graph attention mechanism, SG-SCA leverages the graph structure to capture temporal dependencies in the data, enhancing both the model’s performance and interpretability. Experimental results on the ASCAD-fix dataset with masking protection and the AES-HD dataset with parallel processing protection show that SG-SCA provides concise and accurate interpretations via attention matrix visualization and weighted betweenness centrality analysis. This work offers a new and transparent deep learning model for side-channel analysis, facilitating security evaluation and guiding the design of more robust cryptographic implementations.

As the sole NIST-standardized quantum-resistant key encapsulation mechanism, CRYSTALS-Kyber demands rigorous scrutiny of its side-channel countermeasures. However, there is a lack of research on side-channel security for the message decoding module in masked CRYSTALS-Kyber. In this paper, we seek to address this gap. First, we conduct a side-channel security evaluation of the first-order masked message decoding function in mkm4 of CRYSTALS-Kyber, finding that an incremental storage vulnerability still exists. Then, we implement a practical experiment in the Cortex-M4 CPU using the sum-of-squared difference method, with the accuracy of the message recovery reaching 90.6% and the secret key recovery achieving 77.2%. Furthermore, we theoretically analyze that any order of masking strategy cannot effectively protect the message decoding function, except by increasing the attack difficulty to a limited extent. We also provide our idea for solving this problem by emulating the data behavior of the dual-rail pre-charge logic circuit at the software level, which can effectively ensure the implementation security of CRYSTALS-Kyber.

Side-channel attacks are an effective method for obtaining keys by utilizing signals such as power consumption and electromagnetic radiation generated by cryptographic devices. Intuitively, multiple channels provide richer information than a single channel, suggesting that multi-channel fusion attacks (MCFAs) should outperform single-channel attacks. However, existing MCFAs show only limited improvement and sometimes even perform worse than single-channel attacks. Moreover, current research does not clarify when MCFAs are a better choice. To alleviate this, we use the Hadamard Product Fusion Algorithm (HPFA) proposed in this paper as an example to conduct a thorough analysis and derive the Signal-to-Noise Ratio (SNR) equation for the fused data obtained by HPFA. We find that whether HPFA is superior depends on the correlation coefficient of the noise between channels and the ratio of SNRs of the individual channels. When the noise correlation coefficient is high, HPFA only shows advantages if the SNRs of the single channels are very close to each other. Conversely, when the noise correlation coefficient is low, HPFA can outperform single-channel attacks even if there are significant differences in SNRs. Based on this, we optimize HPFA by incorporating time-frequency decomposition, further enhancing its performance. Experiments validate the theoretical analysis and demonstrate that our methods significantly outperforms existing work when the conditions mentioned earlier are met.

Success Rate (SR) is one of the most popular side-channel security metrics measuring the efficiency of key recovery. Theoretical expression of success rate reveals the functional dependency between relevant parameters such as number of measurements and Signal-to-Noise Ratio (SNR), helping researchers understand the resistance of a given implementation rapidly. However so far, existing works have exposed fundamental problems: (1) Evaluation is confined to a very limited range of distinguishers and specialized methods; (2) Evaluation assumes a perfect leakage model that is detached from reality. It is widely observed that an inaccurate leakage model will lead to a degraded or even distorted success rate. In this paper, we tackle above problems by introducing a novel framework which is able to evaluate seven side-channel distinguishers with a unified expression. Among them, we explore four new distinguishers that have not been investigated in the existing literature. Within the framework, DPA distinguishers are intuitively understood as linear maximum likelihood attack testing closeness between vectors with some easy-to-comprehend geometric metrics. Our evaluation is able to deal with profiled models of any quality and is agnostic to model profiling techniques. It uniquely enables the evaluation of success rates under inaccurate leakage models, whilst providing an (indirect) answer to the open question “how much information is lost due to the model biases” through quantifying the degradation of success rates. Finally, we formulate a set of criterion values for quantitative analyses of the model biases. It provides theoretical evidences for a more thorough explanation for the various behaviors of DPA attacks. Experimental results are inline with the theory, confirming its practical applicability.

: Masking, as a common countermeasure, has been widely utilized to protect cryptographic implementations against power side-channel attacks. It significantly enhances the difficulty of attacks, as the sensitive intermediate values are randomly partitioned into multiple parts and executed on different times. The adversary must amalgamate information across diverse time samples before launching an attack, which is generally accomplished by feature extraction (e.g., Points-Of-Interest (POIs) combination and dimensionality reduction). However, traditional POIs combination methods, machine learning and deep learning techniques are often too time consuming, and necessitate a significant amount of computational resources. In this paper, we undertake the first study on manifold learning and their applications against masked cryptographic implementations. The leaked information, which manifests as the manifold of high-dimensional power traces, is mapped into a low-dimensional space and achieves feature extraction through manifold learning techniques like ISOMAP, Locally Linear Embedding (LLE), and Laplacian Eigenmaps (LE). Moreover, to reduce the complexity, we further construct explicit polynomial mappings for manifold learning to facilitate the dimensionality reduction. Compared to the classical machine learning and deep learning techniques, our schemes built from manifold learning techniques are faster, unsupervised, and only require very simple parameter tuning. Their effectiveness has been fully validated by our detailed experiments.

In this work, we propose practical side-channel attacks for message recovery in post-quantum key encapsulation mechanisms (KEM). As a target scheme, Kyber is a standardized algorithm in the ongoing NIST standardization process. Notably, this work is the first one that implements message recovery by exploiting the information leaked on computational operations during Kyber decoding. The main findings include 1. analyzing computational operations during decoding by power consumption information to effectively recover message; 2. recovering message by analyzing the time differences existing in decoding single bits; 3. by way of simple power analysis, using incremental storage leakage to recover the message.

Aiming at the problem that current schemes mostly let the latter signer to verify the multiple signatures of the previous signer, and the message is signed and then passed to the next signer, leading to the efficiency problem of long overall signature time.A sequential aggregate signature scheme based on bilinear pairs was proposed.The aggregate signatures of documents were generated by multiple users in a certain order, and only the final signature was needed to be verified to confirm the correct order of signatures and the legitimacy of multiple user signatures.The complexity of verifying the multi-user sequential signature was effectively reduced and the offline verification of the authenticity of signature was realized when the user was offline or in a delay-tolerant network with limited node caching capacity and network resources.It is shown that the proposed scheme is existential unforgeability against chosen-message attacks under adversary adaptive selection messages in the random oracle model.

Objective: The popularity of network technology makes more and more enterprises and individuals join the wave of the Internet, and data presents an explosive exponential growth trend.With the increasing demand for data security transmission and fine-grained authentication, the use of cryptographic services in various applications is becoming more frequent. How to deal with random cross and large peak difference cryptographic service requests has gradually become a bottleneck problem restricting various network security applications.A model of cryptographic service scheduling system is proposed to explore the differential dynamic on-demand scheduling of cryptographic service resources. Methods: Optimized entropy method and cryptographic resource reconstruction technology were used to provide dynamic and extensible cryptographic service resources for users and devices accessing service system. Firstly, the evaluation method of cryptographic device service ability is proposed. By obtaining the operating state information such as the utilization rate of cryptographic resources and network throughput of cryptographic devices,the optimized entropy method is used to process the data. Combined with the cryptographic resource allocation of cryptographic devices, the cryptographic service ability provided by cryptographic devices is described,which provides support for cryptographic job scheduling.Then, an efficient on-demand cryptographic job scheduling strategy is proposed, and the cryptographic service request expectation is proposed. By calculating the load distance of the cryptographic device to determine whether to meet the requirements of the cryptographic service, the cryptographic job scheduling strategy is generated. In addition,the cryptographic devices can be reconstructed according to the scheduling algorithm to meet the differentiated needs of cryptographic services in terms of service quality and service efficiency. Results:The enhanced Min-Min load balancing algorithm,the cluster load balancing algorithm based on dynamic consistent hashing and the proposed on-demand scheduling algorithm are used for comparison. By sending cryptographic service requests, the maximum completion time of cryptographic operations, the number of serviceable requests per unit time and the average load of FPGA(field programmable gate array)cryptographic computing unit of the three scheduling algorithms are tested respectively.Fig.7 shows that when the number of cryptographic service requests is small,the difference among the three scheduling algorithms is not obvious.However, with the increase of the number of cryptographic service requests,the load of FPGA computing unit gradually increases. The other two scheduling algorithms do not consider the migration of cryptographic jobs and the dynamic configuration of FPGA computing unit, and the queuing time of cryptographic jobs increases significantly, and the gap between the other two scheduling algorithms and the on-demand scheduling algorithm is getting bigger and bigger.Fig.8 shows that when the number of cryptographic service requests is small, the difference of the three scheduling algorithms is not obvious,which can meet most of the cryptographic service requests. However, with the increase of the number of cryptographic service requests, the number of service requests per unit time of the three scheduling algorithms reaches the peak.Because the on-demand scheduling algorithm realizes the cryptographic job migration and the dynamic configuration of FPGA computing units, the number of service requests per unit time is higher than the other two scheduling algorithms.Fig. 9 shows that under the premise of minimizing the migration of cryptographic operations and the reconstruction of FPGA computing units, the on-demand scheduling algorithm prioritizes the cryptographic operations to the same FPGA computing unit.Therefore,only one FPGA computing unit has load when the number of cryptographic service requests is small, and with the increase of the number of cryptographic service requests, the number of FPGA computing units working also increases. Figs. 10 – 11 show that the FPGA load of the other two algorithms is relatively balanced.When the number of cryptographic service requests is large, the load of each FPGA is high.When the new cryptographic service request arrives,the residual calculation ability of FPGA calculation unit is insufficient to meet the cryptographic service demand because the migration of cryptographic jobs and the dynamic configuration of FPGA calculation unit are not considered. Conclusions: An efficient on-demand scheduling scheme for cryptographic service resources is proposed. The description and dynamic monitoring of cryptographic service capability are realized by using the normalized evaluation model of cryptographic devices based on optimized entropy method. At the same time, a cryptographic job scheduling strategy suitable for different requirements is proposed, and combined with the cryptographic resource reconstruction strategy,the differential configuration and scheduling of cryptographic resources are realized. The dynamic and extensible cryptographic service resources are provided to users and devices of any access service system.

The performance of Side-Channel Attacks (SCAs) decays rapidly when considering more sub-keys, making the full-key recovery a very challenging problem. Limited to independent collision information utilization, collision attacks establish the relationship among sub-keys but do not significantly slow down this trend. To solve it, we first exploit the samples from the previously attacked S-boxes to assist attacks on the targeted S-box under an assumption that similar leakage occurs in program loop or code reuse scenarios. The later considered S-boxes are easier to be recovered since more samples participate in this assist attack, which results in the “snowball” effect. We name this scheme as Snowball, which significantly slows down the attenuation rate of attack performance. We further introduce confusion coefficient into the collision attack to construct collision confusion coefficient, and deduce its relationship with correlation coefficient. Based on this relationship, we give two optimizations on our Snowball exploiting the “values” information and “rankings” information of collision correlation coefficients named Least Deviation from Pearson correlation coefficient (PLD) and Least Deviation from confusion coefficient (CLD). Experiments show that the above optimizations significantly improve the performance of our Snowball.

. Success Rate (SR) is empirically and theoretically a common metric for evaluating the performance of side-channel attacks. Intuitive expressions of success rate are desirable since they reveal and explain the functional dependence on relevant parameters, such as number of measurements and Signal-to-Noise Ratio (SNR), in a straightforward man-ner. Meanwhile, existing works more or less expose unsolved fundamental problems, such as strong leakage assumption, difficulty in interpretation of principle, inaccurate evaluation, and inconsideration of high-order SR. In this paper, we first provide an intuitive framework that statistical tests embedded in different univariate DPA attacks are unified as analyzing and comparing visualized vectors in a Euclidean space by using different easy-to-understand metrics. Then, we establish a unified framework to abstract and convert the security evaluations to the problem of finding a boundary in the Euclidean space. With expressions of the boundary, judging whether a DPA attack succeeds in sense of o th -order becomes fairly efficient and intuitive, and the corresponding SR can be calculated theoretically by integral. Finally, we propose an algorithm that is capable of estimating arbitrary order of SR effectively. Our experimental results verify the theory and highlight the superiority. We believe our research raises many new perspectives for comparing and evaluating side-channel attacks, countermeasures and implementations.

By introducing collision information into side-channel distinguishers, the existing collision-optimized attacks exploit collision detection algorithm to transform the original candidate space under consideration into a significantly smaller collision chain space, thus achieving more efficient key recovery. However, collision information is detected very repeatedly since collision chains are created from the same sub-chains, i.e., with the same candidates on their first several sub-keys. This aggravates when exploiting more collision information. The existing collision detection algorithms try to alleviate this, but the problem is still very serious. In this paper, we propose a highly-efficient detection algorithm named Collision Tree (CoTree) for collision-optimized attacks. CoTree exploits tree structure to store the chains creating from the same sub-chain on the same branch. It then exploits a top-down tree building procedure and traverses each node only once when detecting their collisions with a candidate of the sub-key currently under consideration. Finally, it launches a bottom-up branch removal procedure to remove the chains unsatisfying the collision conditions from the tree after traversing all candidates (within given threshold) of this sub-key, thus avoiding the traversal of the branches satisfying the collision condition. These strategies make our CoTree significantly

Aiming at the privacy leakage,which was caused by collecting user information by third-party host in Android operating system App,a privacy leakage evaluation scheme HostRisk was proposed.HostRisk was based on TF-IDF model and hierarchical clustering method,which was applied in mobile device.The TF-IDF model calculated the business relevance between Apps and hosts via the behavior characteristics of the hosts in these Apps.For the business related hosts that fail to express the business relevance characteristics,those hosts were adjusted and optimized via the average connected hierarchical agglomerative clustering method.Finally,the harmful degree of privacy leakage was evaluated based on the ranking of all hosts in the App.The experimental results verify the effectiveness and efficiency of the scheme.

Factors such as link differences and TCP friendliness constraints lead to the problem of unfair bandwidth allocation when the TCP congestion control mechanism was applied directly to multi-path transmission.To address the problem,it was proposed that a multipath congestion control algorithm was based on link capacity.The proposed algorithm which was based on the concept of feedback regulation achieved multipath joint congestion control by establishing the M/M/1 cache queue model to adjust the throughput rate of senders.Experimental results show the proposed algorithm can improve the multipath transmission bandwidth utilization and the multipath congestion control algorithm responsiveness,and ensure the fairness of multipath transmission.

Key enumeration schemes are used to post-process the scores given by side channel distinguishers and enumerate the key candidates from the most possible one to the least possible one, which can be regarded as optimal tools of key search. However, the application of them is limited by very large key candidate space and computing power consumption. For example, the attacker may spend several weeks or months enumerating the whole 245 key candidates. Unlike the former literature that try to propose a more efficient algorithm to process the distinguishers, scores of key candidates directly, we focus on pre-processing and reducing the key candidate space. To achieve this goal, a new divide and conquer strategy named group collision attack (GCA) is proposed in this paper. The GCA works as follows in brief. The key candidates are first divided into groups on which intra-group collision attack is used to remove the impossible key combinations in each group. Then, the inter-group collision attack is performed to further remove the impossible key combinations between groups. Thus, the complexity of key enumeration is reduced significantly. A series of practical experiments are carried out by using our GCA and the experimental results verify its efficiency.

企业级网络中存在的漏洞日益增多,给公司网络系统安全控制机制的优化选择带来了巨大挑战。本文通过对企业网络中漏洞之间的复杂依赖关系进行建模,构建了漏洞依赖图,并在此基础上建立了Stackelberg攻防博弈模型。同时考虑到传统求解方法无法求解实际的问题规模,引入双模块算法。实验结果表明,本文提出的模型和方法是可行的、高效的。

The rise of e-commerce and other network services have greatly promoted the development of electronic bill service,while traditional data storage schemes can no longer satisfy the rapid access requirements of massive electronic bill data.To solve these problems,a hierarchical scalable storage architecture for massive electronic bill was proposed,which combined hash modular algorithm and consistent hash algorithm,and supported fast data locating.A horizontal expansion scheme based on hash modular algorithm was designed to reduce the amount of data that needed to migrate when adding or deleting data nodes.Besides,a data caching scheme based on hot data and a load balancing scheme based on access delay were designed and implemented,further improving the speed of data access.Finally,the experiments prove the effectiveness of proposed architecture and schemes.

Distinguishers play an important role in Side Channel Analysis (SCA), where real world leakage information is compared against hypothetical predictions in order to guess at the underlying secret key. However, the direct relationship between leakages and predictions can be disrupted by the mathematical combining of d random values with each sensitive intermediate value of the cryptographic algorithm (a so-called “d-th order masking scheme”). In the case of software implementations, as long as the masking has been correctly applied, the guessable intermediates will be independent of any one point in the trace, or indeed of any tuple of fewer than d+1 points. However, certain d+1 -tuples of time points may jointly depend on the guessable intermediates. A typical approach to exploiting this data dependency is to pre-process the trace – computing carefully chosen univariate functions of all possible d+1 -tuples – before applying the usual univariate distinguishers. This has a computational complexity which is exponential in the order d of the masking scheme. In this paper, we propose a new distinguisher based on Kernel Discriminant Analysis (KDA) which directly exploits properties of the mask implementation without the need to exhaustively pre-process the traces, thereby distinguishing the correct key with lower complexity. Experimental results for 2nd and 3rd order attacks (i.e. against 1st and 2nd order masking) verify that the KDA is an effective distinguisher in protected settings.

Side-channel distinguishers play an important role in differential power analysis, where real world leakage information is compared against hypothetical predictions in order to guess at the underlying secret key. A class of distinguishers which can be described as 'cluster-based' have the advantage that they are able to exploit multi-dimensional leakage samples in scenarios where only loose, 'semi-profiled' approximations of the true leakage forms are available. This is by contrast with univariate distinguishers exploiting only single points (e.g. correlation), and Template Attacks requiring concise fitted models which can be overly sensitive to mismatch between the profiling and attack acquisitions. This paper collects together-to our knowledge, for the first time-the various different proposals for cluster-based DPA (concretely, Differential Cluster Analysis, First Principal Components Analysis, and Linear Discriminant Analysis), and shows how they fit within the robust 'semi-profiling' attack procedure proposed by Whitnall et al. at CHES 2015. We provide discussion of the theoretical similarities and differences of the separately proposed distinguishers as well as an empirical comparison of their performance in a range of (real and simulated) leakage scenarios and with varying parameters. Our findings have application for practitioners constrained to rely on 'semi-profiled' models who wish to make informed choices about the best known procedures to exploit such information.