Cryptography and Security (cs.CR)

Abstract: The escalating digitalisation of our lives and enterprises has led to a parallel growth in the complexity and frequency of cyber-attacks. Small and medium-sized enterprises (SMEs), particularly in Australia, are experiencing increased vulnerability to cyber threats, posing a significant challenge to the nation's cyber security landscape. Embracing transformative technologies such as Artificial Intelligence (AI), Machine Learning (ML) and Large Language Models (LLMs) can potentially strengthen cyber security policies for Australian SMEs. However, their practical application, advantages, and limitations remain underexplored, with prior research mainly focusing on large corporations. This study aims to address this gap by providing a comprehensive understanding of the potential role of LLMs in enhancing cyber security policies for Australian SMEs. Employing a mixed-methods study design, this research includes a literature review, qualitative analysis of SME case studies, and a quantitative assessment of LLM performance metrics in cyber security applications. The findings highlight the promising potential of LLMs across various performance criteria, including relevance, accuracy, and applicability, though gaps remain in areas such as completeness and clarity. The study underlines the importance of integrating human expertise with LLM technology and refining model development to address these limitations. By proposing a robust conceptual framework guiding the effective adoption of LLMs, this research aims to contribute to a safer and more resilient cyber environment for Australian SMEs, enabling sustainable growth and competitiveness in the digital era.

Abstract: Mixed Integer Linear Programming (MILP) is a well-known approach for the cryptanalysis of a symmetric cipher. A number of MILP-based security analyses have been reported for non-linear (SBoxes) and linear layers. Researchers proposed word- and bit-wise SBox modeling techniques using a set of inequalities which helps in searching differential trails for a cipher. In this paper, we propose two new techniques to reduce the number of inequalities to represent the valid differential transitions for SBoxes. Our first technique chooses the best greedy solution with a random tiebreaker and achieves improved results for the 4-bit SBoxes of MIBS, LBlock, and Serpent over the existing results of Sun et al. [25]. Subset addition, our second approach, is an improvement over the algorithm proposed by Boura and Coggia. Subset addition technique is faster than Boura and Coggia [10] and also improves the count of inequalities. Our algorithm emulates the existing results for the 4-bit SBoxes of Minalpher, LBlock, Serpent, Prince, and Rectangle. The subset addition method also works for 5-bit and 6-bit SBoxes. We improve the boundary of minimum number inequalities from the existing results for 5-bit SBoxes of ASCON and SC2000. Application of subset addition technique for 6-bit SBoxes of APN, FIDES, and SC2000 enhances the existing results. By applying multithreading, we reduced the execution time needed to find the minimum inequality set over the existing techniques.

Abstract: The vast increase of IoT technologies and the ever-evolving attack vectors and threat actors have increased cyber-security risks dramatically. Novel attacks can compromise IoT devices to gain access to sensitive data or control them to deploy further malicious activities. The detection of novel attacks often relies upon AI solutions. A common approach to implementing AI-based IDS in distributed IoT systems is in a centralised manner. However, this approach may violate data privacy and secrecy. In addition, centralised data collection prohibits the scale-up of IDSs. Therefore, intrusion detection solutions in IoT ecosystems need to move towards a decentralised direction. FL has attracted significant interest in recent years due to its ability to perform collaborative learning while preserving data confidentiality and locality. Nevertheless, most FL-based IDS for IoT systems are designed under unrealistic data distribution conditions. To that end, we design an experiment representative of the real world and evaluate the performance of two FL IDS implementations, one based on DNNs and another on our previous work on DBNs. For our experiments, we rely on TON-IoT, a realistic IoT network traffic dataset, associating each IP address with a single FL client. Additionally, we explore pre-training and investigate various aggregation methods to mitigate the impact of data heterogeneity. Lastly, we benchmark our approach against a centralised solution. The comparison shows that the heterogeneous nature of the data has a considerable negative impact on the model performance when trained in a distributed manner. However, in the case of a pre-trained initial global FL model, we demonstrate a performance improvement of over 20% (F1-score) when compared against a randomly initiated global model.

Abstract: The rapid expansion of the use of blockchain-based systems often leads to a choice between customizable private blockchains and more secure, scalable and decentralized but expensive public blockchains. This choice represents the trade-off between privacy and customization at a low cost and security, scalability, and a large user base but at a high cost. In order to improve the scalability of secure public blockchains while enabling privacy and cost reduction, zk-rollups, a layer 2 solution, appear to be a promising avenue. This paper explores the benefits of zk-rollups, including improved privacy, as well as their potential to support transactions designed for specific applications. We propose an innovative design that allows multiple zk-rollups to co-exist on the same smart contracts, simplifying their creation and customization. We then evaluate the first implementation of our system highlighting a low overhead on existing transaction types and on proof generation while strongly decreasing the cost of new transaction types and drastically reducing zk-rollup creation costs.

Abstract: Decision-based evasion attacks repeatedly query a black-box classifier to generate adversarial examples. Prior work measures the cost of such attacks by the total number of queries made to the classifier. We argue this metric is flawed. Most security-critical machine learning systems aim to weed out "bad" data (e.g., malware, harmful content, etc). Queries to such systems carry a fundamentally asymmetric cost: queries detected as "bad" come at a higher cost because they trigger additional security filters, e.g., usage throttling or account suspension. Yet, we find that existing decision-based attacks issue a large number of "bad" queries, which likely renders them ineffective against security-critical systems. We then design new attacks that reduce the number of bad queries by $1.5$-$7.3\times$, but often at a significant increase in total (non-bad) queries. We thus pose it as an open problem to build black-box attacks that are more effective under realistic cost metrics.

Abstract: Malicious server (MS) attacks have enabled the scaling of data stealing in federated learning to large batch sizes and secure aggregation, settings previously considered private. However, many concerns regarding client-side detectability of MS attacks were raised, questioning their practicality once they are publicly known. In this work, for the first time, we thoroughly study the problem of client-side detectability.We demonstrate that most prior MS attacks, which fundamentally rely on one of two key principles, are detectable by principled client-side checks. Further, we formulate desiderata for practical MS attacks and propose SEER, a novel attack framework that satisfies all desiderata, while stealing user data from gradients of realistic networks, even for large batch sizes (up to 512 in our experiments) and under secure aggregation. The key insight of SEER is the use of a secret decoder, which is jointly trained with the shared model. Our work represents a promising first step towards more principled treatment of MS attacks, paving the way for realistic data stealing that can compromise user privacy in real-world deployments.

Abstract: The remarkable proliferation of deep learning across various industries has underscored the importance of data privacy and security in AI pipelines. As the evolution of sophisticated Membership Inference Attacks (MIAs) threatens the secrecy of individual-specific information used for training deep learning models, Differential Privacy (DP) raises as one of the most utilized techniques to protect models against malicious attacks. However, despite its proven theoretical properties, DP can significantly hamper model performance and increase training time, turning its use impractical in real-world scenarios. Tackling this issue, we present Discriminative Adversarial Privacy (DAP), a novel learning technique designed to address the limitations of DP by achieving a balance between model performance, speed, and privacy. DAP relies on adversarial training based on a novel loss function able to minimise the prediction error while maximising the MIA's error. In addition, we introduce a novel metric named Accuracy Over Privacy (AOP) to capture the performance-privacy trade-off. Finally, to validate our claims, we compare DAP with diverse DP scenarios, providing an analysis of the results from performance, time, and privacy preservation perspectives.