This paper considers the security of CRAFT and WARP. We present a practical key-recovery attack on full-round CRAFT in the related-key setting with only one differential characteristic, and the theoretical time complexity of the attack is $2^{36.09}$ full-round encryptions. The attack is verified in practice. The test result indicates that the theoretical analysis is valid, and it takes about $15.69$ hours to retrieve the key. A full-round key-recovery attack on WARP in the related-key setting is proposed, and the time complexity is $2^{44.58}$ full-round encryptions. The theoretical attack is implemented on a round-reduced version of WARP, which guarantees validity. Besides, we give a 33-round multiple zero-correlation linear attack on WARP, which is the longest attack on the cipher in the single-key attack setting. We note that the attack results in this paper do not threaten the security of CRAFT and WARP as the designers do not claim security under the related-key attack setting.

In this paper, we outline a novel form of attack we refer to as Opportunistic Algorithmic Double-Spending (OpAl ). OpAl attacks avoid equivocation, i.e., do not require conflicting transactions, and are carried out automatically in case of a fork. Algorithmic double-spending is facilitated through transaction semantics that dynamically depend on the context and ledger state at the time of execution. Hence, OpAl evades common double-spending detection mechanisms and can opportunistically leverage forks, even if the malicious sender themselves is not responsible for, or even actively aware of, any fork. Forkable ledger designs with expressive transaction semantics, especially stateful EVM-based smart contract platforms such as Ethereum, are particularly vulnerable. Hereby, the cost of modifying a regular transaction to opportunistically perform an OpAl attack is low enough to consider it a viable default strategy. While Bitcoin’s stateless UTXO model, or Cardano’s EUTXO model, appear more robust against OpAl , we nevertheless demonstrate scenarios where transactions are semantically malleable and thus vulnerable. To determine whether OpAl -like semantics can be observed in practice, we analyze the execution traces of 922562 transactions on the Ethereum blockchain. Hereby, we are able to identify transactions, which may be associated with frontrunning and MEV bots, that exhibit some of the design patterns also employed as part of the herein presented attack.

The Learning with Errors (LWE) problem receives much attention in cryptography, mainly due to its fundamental significance in post-quantum cryptography. Among its solving algorithms, the Blum-Kalai-Wasserman (BKW) algorithm, originally proposed for solving the Learning Parity with Noise (LPN) problem, performs well, especially for certain parameter settings with cryptographic importance. The BKW algorithm consists of two phases, the reduction phase and the solving phase. In this work, we study the performance of distinguishers used in the solving phase. We show that the Fast Fourier Transform (FFT) distinguisher from Eurocrypt’15 has the same sample complexity as the optimal distinguisher, when making the same number of hypotheses. We also show via simulation that it performs much better than previous theory predicts and develop a sample complexity model that matches the simulations better. We also introduce an improved, pruned version of the FFT distinguisher. Finally, we indicate, via extensive experiments, that the sample dependency due to both LF2 and sample amplification is limited.

Aaram Yun et al. considered that Lai-Massey structure has the same security as Feistel structure. However, Luo et al. showed that 3-round Lai-Massey structure can resist quantum attacks of Simon's algorithm, which is different from Feistel structure. We give quantum attacks against a typical Lai-Massey structure. The result shows that there exists a quantum CPA distinguisher against 3-round Lai-Massey structure and a quantum CCA distinguisher against 4-round Lai-Massey Structure, which is the same as Feistel structure. We extend the attack on Lai-Massey structure to quasi-Feistel structure. We show that if the combiner of quasi-Feistel structure is linear, there exists a quantum CPA distinguisher against 3-round balanced quasi-Feistel structure and a quantum CCA distinguisher against 4-round balanced quasi-Feistel Structure.

Misuse resilience is an important security criterion in the evaluation of the NIST Post-quantum cryptography standardization process. In this paper, we propose new key mismatch attacks against Kyber and Saber, NIST's selected scheme for encryption and one of the finalists in the third round of the NIST competition, respectively. Our novel idea is to recover partial information of multiple secret entries in each mismatch oracle call. These multi-positional attacks greatly reduce the expected number of oracle calls needed to fully recover the secret key. They also have significance in side-channel analysis. From the perspective of lower bounds, our new attacks falsify the Huffman bounds proposed in [Qin et al. ASIACRYPT 2021], where a one- positional mismatch adversary is assumed. Our new attacks can be bounded by the Shannon lower bounds, i.e., the entropy of the distribution generating each secret coefficient times the number of secret entries. We call the new attacks "near-optimal" since their query complexities are close to the Shannon lower bounds.

This paper presents a fast method to compute algebraic norms of integral elements of smooth-degree cyclotomic fields, and, more generally, smooth-degree Galois number fields with commutative Galois groups. The typical scenario arising in $S$-unit searches (for, e.g., class-group computation) is computing a $\Theta(n\log n)$-bit norm of an element of weight $n^{1/2+o(1)}$ in a degree-$n$ field; this method then uses $n(\log n)^{3+o(1)}$ bit operations. An $n(\log n)^{O(1)}$ operation count was already known in two easier special cases: norms from power-of-2 cyclotomic fields via towers of power-of-2 cyclotomic subfields, and norms from multiquadratic fields via towers of multiquadratic subfields. This paper handles more general Abelian fields by identifying tower-compatible integral bases supporting fast multiplication; in particular, there is a synergy between tower-compatible Gauss-period integral bases and a fast-multiplication idea from Rader. As a baseline, this paper also analyzes various standard norm-computation techniques that apply to arbitrary number fields, concluding that all of these techniques use at least $n^2(\log n)^{2+o(1)}$ bit operations in the same scenario, even with fast subroutines for continued fractions and for complex FFTs. Compared to this baseline, algorithms dedicated to smooth-degree Abelian fields find each norm $n/(\log n)^{1+o(1)}$ times faster, and finish norm computations inside $S$-unit searches $n^2/(\log n)^{1+o(1)}$ times faster.

In this paper, we aim to present a quantum setting oriented preimage attack against 4-round Keccak-224. An important technique we called the allocating rotational cryptanalysis takes the preimage attack into the situation of 2-block preimage recovery. With the conditions on the middle state proposed by Li et al., we use the generic quantum preimage attack to deal with the finding of first preimage block. By using the newly explored propagation of rotational relations, we significantly increase the number of eigenpoints at the end of 4-round modified Keccak-f from 0 to 32, and therefore improving the accuracy of determining the rotational number for a certain rotational counterpart in the quantum setting by more than 10 orders of magnitude. On the basis of the above, we design an efficient unitary oracle operator with only twice calling of the 4-round modified Keccak-f, which costs half of previous results, to mark a rotational counterpart of the second preimage block in order that the second preimage block can be found indirectly from a quickly generated specified search space. As a result on the 4-round Keccak-224: In the classical setting, the preimage attack with the complexity decreased to 2^218 is better than the result based on the pioneered rotational cryptanalysis. In the quantum setting, the amplitude amplification driven preimage attack with a complexity of 2^110 is by far the best dedicated quantum preimage attack. Additionally, the SKW algorithm is applied to the dedicated quantum preimage attack against the 4-round Keccak-224 for the first time, which is exponentially easier to implement in quantum circuit than the former, with a complexity of 2^111.

The adoption of deep neural networks for profiling side-channel attacks (SCA) opened new perspectives for leakage detection. Recent publications showed that cryptographic implementations featuring different countermeasures could be broken without feature selection or trace preprocessing. This success comes with a high price: extensive hyperparameter search to find optimal deep learning models. As deep learning models usually suffer from overfitting due to their high fitting capacity, it is crucial to avoid over-training regimes, which require a correct number of epochs. For that, \textit{early stopping} is employed as an efficient regularization method that requires a consistent validation metric. Although guessing entropy is a highly informative metric for profiling SCA, it is time-consuming, especially if computed for all epochs during training and the number of validation traces is significantly large. This paper shows that guessing entropy can be efficiently computed during training by reducing the number of validation traces without affecting the efficiency of early stopping decisions. Our solution significantly speeds up the process, impacting hyperparameter search and overall profiling attack performances. Our fast guessing entropy calculation is up to 16$\times$ faster, resulting in more hyperparameter tuning experiments and allowing security evaluators to find more efficient deep learning model.

The deep learning-based side-channel analysis gave some of the most prominent side-channel attacks against protected targets in the past few years. To this end, the research community's focus has been on creating 1) powerful and 2) (if possible) minimal multilayer perceptron or convolutional neural network architectures. Currently, we see that computationally intensive hyperparameter tuning methods (e.g., Bayesian optimization or reinforcement learning) provide the best results. However, as targets with more complex countermeasures become available, these minimal architectures may be insufficient, and we will require novel deep learning approaches. This work explores how residual neural networks (ResNets) perform in side-channel analysis and how to construct deeper ResNets capable of working with larger input sizes and requiring minimal tuning. The resulting architectures obtained by following our guidelines are significantly deeper than commonly seen in side-channel analysis, require minimal hyperparameter tuning for specific datasets, and offer competitive performance with state-of-the-art methods across several datasets. Additionally, the results indicate that ResNets work especially well when the number of profiling traces and features in a trace is large.

MEGA is a leading cloud storage platform with more than 250 million users and 1000 Petabytes of stored data. MEGA claims to offer user-controlled, end-to-end security. This is achieved by having all data encryption and decryption operations done on MEGA clients, under the control of keys that are only available to those clients. This is intended to protect MEGA users from attacks by MEGA itself, or by adversaries who have taken control of MEGA’s infrastructure. We provide a detailed analysis of MEGA’s use of cryptography in such a malicious server setting. We present five distinct attacks against MEGA, which together allow for a full compromise of the confidentiality of user files. Additionally, the integrity of user data is damaged to the extent that an attacker can insert malicious files of their choice which pass all authenticity checks of the client. We built proof-of-concept versions of all the attacks. Four of the five attacks are eminently practical. They have all been responsibly disclosed to MEGA and remediation is underway. Taken together, our attacks highlight significant shortcomings in MEGA’s cryptographic architecture. We present immediately deployable countermeasures, as well as longer-term recommendations. We also provide a broader discussion of the challenges of cryptographic deployment at massive scale under strong threat models.

In this work, we recover the private key material of the FrodoKEM key exchange mechanism as submitted to the NIST Post Quantum Cryptography (PQC) standardization process. The new mechanism that allows for this is a Rowhammer-assisted \emph{poisoning} of the FrodoKEM Key Generation (KeyGen) process. The Rowhammer side-channel is a hardware-based security exploit that allows flipping bits in DRAM by “hammering” rows of memory adjacent to some target-victim memory location by repeated memory accesses. Using Rowhammer, we induce the FrodoKEM software to output a higher-error Public Key (PK), $(\mathbf{A}, \mathbf{B} = \mathbf{A}\mathbf{S}+\mathbf{\widetilde{E}}),$ where the error $\widetilde{\mathbf{E}}$ is modified by Rowhammer. Then, we perform a decryption failure attack, using a variety of publicly-accessible supercomputing resources running on the order of only 200,000 core-hours. We delicately attenuate the decryption failure rate to ensure that the adversary's attack succeeds practically, but so honest users cannot easily detect the manipulation. Achieving this public key "poisoning" requires an extreme engineering effort, as FrodoKEM's KeyGen runs on the order of 8 milliseconds. (Prior Rowhammer-assisted attacks against cryptography require as long as 8 hours of persistent access.) In order to handle this real-world timing condition, we require a wide variety of prior and brand new, low-level engineering techniques, including e.g. memory massaging algorithms -- i.e. "Feng Shui" -- and a precisely-targeted performance degradation attack on the extendable output function SHAKE. We explore the applicability of our techniques to other lattice-based KEMs in the NIST PQC Round 3 candidate-pool, e.g. Kyber, Saber, etc, as well as the difficulties that arise in the various settings. To conclude, we discuss various simple countermeasures to protect implementations against this, and similar, attacks.

Searchable Encryption schemes provide secure search over encrypted databases while allowing admitted information leakages. Generally, the leakages can be categorized into access and volume pattern. In most existing SE schemes, these leakages are caused by practical designs but are considered an acceptable price to achieve high search efficiency. Recent attacks have shown that such leakages could be easily exploited to retrieve the underlying keywords for search queries. Under the umbrella of attacking SE, we design a new Volume and Access Pattern Leakage-Abuse Attack (VAL-Attack) that improves the matching technique of LEAP (CCS ’21) and exploits both the access and volume patterns. Our proposed attack only leverages leaked documents and the keywords present in those documents as auxiliary knowledge and can effectively retrieve document and keyword matches from leaked data. Furthermore, the recovery performs without false positives. We further compare VAL-Attack with two recent well-defined attacks on several real-world datasets to highlight the effectiveness of our attack and present the performance under popular countermeasures.

This work presents new advances in algebraic cryptanalysis of small scale derivatives of AES. We model the cipher as a system of polynomial equations over GF(2), which involves only the variables of the initial key, and we subsequently attempt to solve this system using Gröbner bases. We show, for example, that one of the attacks can recover the secret key for one round of AES-128 under one minute on a contemporary CPU. This attack requires only two known plaintexts and their corresponding ciphertexts. We also compare the performance of Gröbner bases to a SAT solver, and provide an insight into the propagation of diffusion within the cipher.

In this paper, we present the first side-channel attack on a higher-order masked implementation of an IND-CCA secure lattice-based key encapsulation mechanism (KEM). Our attack exploits a vulnerability in the procedure for the arithmetic to Boolean conversion which we discovered. On the example of Saber KEM, we demonstrate successful message and secret key recovery attacks on the second- and third-order masked implementations running on a different device than the profiling one. In our experiments, we use the latest publicly available higher-order masked implementation of Saber KEM in which all known vulnerabilities are patched. The presented approach is not specific to Saber and can be potentially applied to other lattice-based PKE and KEM algorithms, including CRYSTALS-Kyber which has been recently selected for standardization by NIST.

Side-channel attacks are powerful non-invasive attacks on cryptographic algorithms. Among such attacks, profiling attacks have a prominent place as they assume an attacker with access to a copy of the device under attack. The attacker uses the device's copy to learn as much as possible about the device and then mount the attack on the target device. In the last few years, Machine Learning has been successfully used in profiling attacks, as such techniques proved to be capable of breaking implementations protected with countermeasures. In the deep learning-based profiling attack, a core problem is finding efficient neural network architectures to evaluate an implementation's security correctly. Unfortunately, this process is time-consuming, and a different neural network configuration usually needs to be defined for every target. Hence, we propose the following process: train a separate autoencoder for each dataset obtained from different cryptographic implementations and devices to receive an encoded version for each one. After that, define a universal model that can break multiple (encoded) datasets. Thus, instead of finding dataset-specific neural network architectures, we reduce the effort to find autoencoders to encode the datasets and a single neural network to break~them.

To defend against the rising threat of quantum computers, NIST initiated their Post-Quantum Cryptography(PQC) standardization process in 2016. During the PQC process, the security against side-channel attacks has received much attention. Lattice-based schemes are considered to be the most promising group to be standardized. Message encoding in lattice-based schemes has been proven to be vulnerable to side-channel attacks, and a first-order masked message encoder has been presented. However, there is still a lack of security evaluation for the first-order masked message encoder under different implementations. In this paper, we analyzed the security of the first-order masked message encoder of Kyber. We found although masked Kyber certainly is able to defend against the previous side-channel attacks, there still exist some exploitable leakages. With the help of the leakages, we proposed a deep learning-based key recovery attack on message encoding of masked Kyber. Our method can recover the original message from masked message encoding and then enable a chosen-ciphertext attack to recover the secret key. In our experiments, the whole secret key of masked Kyber768 was recovered with only 9 traces and the success rate of attack was close to 100%.

Statistical Ineffective Fault Attack (SIFA) has been a threat for implementa-tions of symmetric cryptographic primitives. Unlike Differential Fault At-tacks (DFA) which takes both correct and faulty ciphertexts, SIFA can re-cover the secret key with only correct ciphertexts. The classic SIFA is only effective on fault models with non-uniform distribution of intermediate val-ue. In this paper, we present a new fault model named adjacent-byte model, which describes a non-uniform distribution of relationship between two bytes (i.e. exclusive-or). To the best of our knowledge, it is the first time that this fault model has been proposed. We also show that the adjacent-byte faults can be induced by different fault sources and easy to reproduce. Then a new SIFA attack method called AB-SIFA on symmetric cryptography is proposed. We demonstrate the effectiveness of this new attack by simulating the attack. Finally, our attacks are applied to a software implementations of AES-128 with redundant countermeasure and a hardware AES co-processor, utilizing voltage glitches and clock glitches.

Creating a good deep learning (DL) model is an art which requires expertise in DL and a large set of labeled data for training neural networks. Neither is readily available. In this paper, we introduce a method which enables us to achieve good results with bad DL models. We use simple multilayer perceptron (MLP) networks, trained on a small dataset, which make strongly biased predictions if used without the proposed method. The core idea is to extend the attack dataset so that at least one of its traces has the ground truth label to which the models are biased towards. The effectiveness of the presented method is demonstrated by attacking an ARM Cortex-M4 CPU implementation of Saber KEM, a finalist of the NIST post-quantum cryptography standardization project, on a nRF52832 system-on-chip supporting Bluetooth 5, using amplitude-modulated EM emanations. Previous amplitude-modulated EM emanation-based attacks on Saber KEM could not recover its messages with a sufficiently high probability. We recover messages with the probability 1 from the profiling device and with the probability 0.74 from a different device. Using messages recovered from chosen ciphertexts, we extract the secret key of Saber KEM.

This paper introduces structure to key, in the related-key attack settings. While the idea of structure has been long used in keyrecovery attacks against block ciphers to enjoy the birthday effect, the same had not been applied to key materials due to the fact that key structure results in uncontrolled differences in key and hence affects the validity or probabilities of the differential trails. We apply this simple idea to improve the related-key boomerang attack against AES-256 by Biryukov and Khovratovich in 2009. Surprisingly, it turns out to be effective, i.e., both data and time complexities are reduced by a factor of about 2^8, to 2^92 and 2^91 respectively, at the cost of the amount of required keys increased from 4 to 2^19. There exist some tradeoffs between the data/time complexity and the number of keys. To the best of our knowledge, this is the first essential improvement of the attack against the full AES-256 since 2009. It will be interesting to see if the structure technique can be applied to other AES-like block ciphers, and to tweaks rather than keys of tweakable block ciphers so the amount of required keys of the attack will not be affected.

The selection of secure parameter sets requires an estimation of the attack cost to break the respective cryptographic scheme instantiated under these parameters. The current NIST standardization process for post-quantum schemes makes this an urgent task, especially considering the announcement to select final candidates by the end of 2021. For code-based schemes, recent estimates seemed to contradict the claimed security of most proposals, leading to a certain doubt about the correctness of those estimates. Furthermore, none of the available estimates include most recent algorithmic improvements on decoding linear codes, which are based on information set decoding (ISD) in combination with nearest neighbor search. In this work we observe that all major ISD improvements are build on nearest neighbor search, explicitly or implicitly. This allows us to derive a framework from which we obtain practical variants of all relevant ISD algorithms including the most recent improvements. We derive formulas for the practical attack costs and make those online available in an easy to use estimator tool written in python and C. Eventually, we provide classical and quantum estimates for the bit security of all parameter sets of current code-based NIST proposals.

Ghosh, Kamara and Tamassia (ASIA CCS 2021) presented a Graph Encryption Scheme supporting shortest path queries. We show how to perform a query recovery attack against this GKT scheme when the adversary is given the original graph together with the leakage of certain subsets of queries. Our attack falls within the security model used by Ghosh et al., and is the first targeting schemes supporting shortest path queries. Our attack uses classical graph algorithms to compute the canonical names of the single-destination shortest path spanning trees of the underlying graph and uses these canonical names to pre-compute the set of candidate queries that match each response. Then, when all shortest path queries to a single node have been observed, the canonical names for the corresponding query tree are computed and the responses are matched to the candidate queries from the offline phase. The output is guaranteed to contain the correct query. For a graph on $n$ vertices, our attack runs in time $O(n^3)$ and matches the time complexity of the GKT scheme's setup. We evaluate the attack's performance using the real world datasets used in the original paper and on random graphs, and show that for the real-world datasets as many as 21.9% of the queries can be uniquely recovered and as many as 50% of the queries result in sets of only three candidates.

In a so-called partial key exposure attack one obtains some information about the secret key, e.g. via some side-channel leakage. This information might be a certain fraction of the secret key bits (erasure model) or some erroneous version of the secret key (error model). The goal is to recover the secret key from the leaked information. There is a common belief that, as opposed to e.g. the RSA cryptosystem, most post-quantum cryptosystems are usually resistant against partial key exposure attacks. We strongly question this belief by constructing partial key exposure attacks on code-based, multivariate, and lattice-based schemes (BIKE, Rainbow and NTRU). Our attacks exploit the redundancy that modern PQ cryptosystems inherently use for efficiency reasons. The application and development of techniques from information set decoding plays a crucial role for achieving our results. On the theoretical side, we show non-trivial information leakage bounds that allow for a polynomial time key recovery attack. As an example, for all schemes the knowledge of a constant fraction of the secret key bits suffices to reconstruct the full key in polynomial time. Even if we no longer insist on polynomial time attacks, most of our attacks extend well and remain feasible up to large erasure and error rates. In the case of BIKE for example we obtain attack complexities around 60 bits when half of the secret key bits are erased, or a quarter of the secret key bits are faulty. Our results show that even highly error-prone key leakage of modern PQ cryptosystems may lead to full secret key recoveries.

In this work, we present the first fault injection analysis of the Number Theoretic Transform (NTT). The NTT is an integral computation unit, widely used for polynomial multiplication in several structured lattice-based key encapsulation mechanisms (KEMs) and digital signature schemes. We identify a critical single fault vulnerability in the NTT, which severely reduces the entropy of its output. This in turn enables us to perform a wide-range of attacks applicable to lattice-based KEMs as well as signature schemes. In particular, we demonstrate novel key recovery and message recovery attacks targeting the key generation and encryption procedure of Kyber KEM. We also propose novel existential forgery attacks targeting deterministic and probabilistic signing procedure of Dilithium, followed by a novel verification bypass attack targeting its verification procedure. All proposed exploits are demonstrated with high success rate using electromagnetic fault injection on state-of-the-art implementations of Kyber and Dilithium, from the open-source pqm4 library on the ARM Cortex-M4 microcontroller.

In the ongoing last round of NIST’s post-quantum cryptography standardization competition, side-channel analysis of finalists is a main focus of attention. While their resistance to timing, power and near field electromagnetic (EM) side-channels has been thoroughly investigated, amplitude-modulated EM emanations has not been considered so far. The attacks based on amplitude-modulated EM emanations are more stealthy because they exploit side-channels intertwined into the signal transmitted by an on-chip antenna. Thus, they can be mounted on a distance from the device under attack. In this paper, we present the first results of an amplitude-modulated EM side-channel analysis of one of the NIST PQ finalists, Saber key encapsulation mechanism (KEM), implemented on the nRF52832 (ARM Cortex-M4) system-on-chip supporting Bluetooth 5. By capturing amplitude-modulated EM emanations during decapsulation, we can recover each bit of the session key with 0.91 probability on average.

One of the main challenges cryptography needs to deal with is balancing the performances of a cryptographic primitive with its security. That is why in 2015, the National Institute of Standards and Technologies (NIST) has begun a standardization process to solicit the creation of new lightweight cryptographic algorithms. We then wondered which of this standardization finalists would suit the best to a white-box implementation. To this end, we studied different algorithms structures on their encodability to later develop our white-box encoding solution. Afterwards, we reviewed the standardization finalists on the applicability of our solution to those algorithms, and finally apply it to GIFT, the permutation of GIFT-COFB.

Physical Unclonable Functions (PUFs) have been increasingly used as an alternative to non-volatile memory for the storage of cryptographic secrets. Research on side channel and fault attacks with the goal of extracting these secrets has begun to gain interest but no fault injection attack targeting the necessary error correction within a PUF device has been shown so far. This work demonstrates one such attack on a hardware fuzzy commitment scheme implementation and thus shows a new potential attack threat existing in current PUF key storage systems. After presenting evidence for the overall viability of the profiled attack by performing it on an FPGA implementation, countermeasures are analysed: we discuss the efficacy of hashing helper data with the PUF-derived key to prevent the attack as well as codeword masking, a countermeasure effective against a side channel attack. The analysis shows the limits of these approaches. First, we demonstrate the criticality of timing in codeword masking by confirming the attack's effectiveness on ostensibly protected hardware. Second, our work shows a successful attack without helper data manipulation and thus the potential for sidestepping helper data hashing countermeasures.

The rebound attack was introduced by Mendel et al. at FSE 2009 to fulfill a heavy middle round of a differential path for free, utilizing the degree of freedom from states. The inbound phase was extended to 2 rounds by the Super-Sbox technique invented by Lamberger et al. at ASIACRYPT 2009 and Gilbert and Peyrin at FSE 2010. In ASIACRYPT 2010, Sasaki et al. further reduced the requirement of memory by introducing the non-full-active Super-Sbox. In this paper, we further develop this line of research by introducing Super-Inbound, which is able to connect multiple 1-round or 2-round (non-full-active) Super-Sbox inbound phases by utilizing fully the degrees of freedom from both states and key, yet without the use of large memory. This essentially extends the inbound phase by up to 3 rounds. We applied this technique to find classic or quantum collisions on several AES-like hash functions, and improved the attacked round number by 1 to 5 in targets including AES-128 and SKINNY hashing modes, Saturnin-Hash, and Grostl-512. To demonstrate the correctness of our attacks, the semi-free-start collision on 6-round AES-128-MMO/MP with estimated time complexity $2^{24}$ in classical setting was implemented and an example pair was found instantly on a standard PC.

Deep learning (DL) has revolutionized Side Channel Analysis (SCA) in recent years. One of the major advantages of DL in the context of SCA is that it can automatically handle masking and desynchronization countermeasures, even while they are applied simultaneously for a cryptographic implementation. However, the success of the attack strongly depends on the DL model used for the attack. Traditionally, Convolutional Neural Networks (CNNs) have been utilized in this regard. This work proposes to use Transformer Network (TN) for attacking implementations secured with masking and desynchronization. Our choice is motivated by the fact that TN is good at capturing the dependencies among distant points of interest in a power trace. Furthermore, we show that TN can be made shift-invariant which is an important property required to handle desynchronized traces. Experimental validation on several public datasets establishes that our proposed TN-based model, called TransNet, outperforms the present state-of-the-art on several occasions. Specifically, TransNet outperforms the other methods by a wide margin when the traces are highly desynchronized. Additionally, TransNet shows good attack performance against implementations with desynchronized traces even when it is trained on synchronized traces. The Tensorflow implementation of TransNet is available at https://github.com/suvadeep-iitb/TransNet.

Timing attack is a class of side-channel attacks that aims to leak secret information based on the time it takes to perform different operations. The biggest advantage of a timing attack is that it does not require sophisticated or expensive equipment to be carried out. Side Channels on FHE schemes have been reported on the client side which has the secret key. But the present project aims to delve into the counter intuitive question, can an analysis be performed on the server end which ideally has no information of the secret key. In this report, we investigate when homomorphic operations are performed on the ciphertexts stored in the server, can timing reveal information of the error used to mask the ciphertexts? Finally, can this be utilized to determine the secret key of the ciphering technique?

The LowMC family of block ciphers was first proposed by Albrecht et al. in [ARS+15], specifically targeting adoption in FHE and MPC applications due to its low multiplicative complexity. The construction operates a 3-bit S-box as the sole non-linear transformation in the algorithm. In contrast, both the linear layer and round key generation are achieved through multiplications of full rank matrices over GF(2). The cipher is instantiable using a diverse set of default configurations, some of which have partial non-linear layers i.e., in which the S-boxes are not applied over the entire internal state of the cipher. The significance of cryptanalysing LowMC was elevated by its inclusion into the NIST PQC digital signature scheme PICNIC in which a successful key recovery using a single plaintext/ciphertext pair is akin to retrieving the secret signing key. The current state-of-the-art attack in this setting is due to Dinur [Din21a], in which a novel way of enumerating the roots of a Boolean system of equation is morphed into a key recovery procedure that undercuts an ordinary exhaustive search in terms of time complexity for the variants of the cipher up to five rounds. In this work, we demonstrate that this technique can efficiently be enriched with a specific linearization strategy that reduces the algebraic degree of the non-linear layer as put forward by Banik et al. [BBDV20]. This amalgamation yields a drastic reduction in terms of memory complexity across all instantiations of LowMC up to six rounds with a quasi-equivalent time complexity.