Federated Learning (FL) has recently arisen as a revolutionary approach to collaborative training Machine Learning models. According to this novel framework, multiple participants train a global model collaboratively, coordinating with a central aggregator without sharing their local data. As FL gains popularity in diverse domains, security, and privacy concerns arise due to the distributed nature of this solution. Therefore, integrating this strategy with Blockchain technology has been consolidated as a preferred choice to ensure the privacy and security of participants. This paper explores the research efforts carried out by the scientific community to define privacy solutions in scenarios adopting Blockchain-Enabled FL. It comprehensively summarizes the background related to FL and Blockchain, evaluates existing architectures for their integration, and the primary attacks and possible countermeasures to guarantee privacy in this setting. Finally, it reviews the main application scenarios where Blockchain-Enabled FL approaches have been proficiently applied. This survey can help academia and industry practitioners understand which theories and techniques exist to improve the performance of FL through Blockchain to preserve privacy and which are the main challenges and future directions in this novel and still under-explored context. We believe this work provides a novel contribution concerning the previous surveys and is a valuable tool to explore the current landscape, understand perspectives, and pave the way for advancements or improvements in this amalgamation of Blockchain and Federated Learning.

@en

Large-scale online campaigns, malicious or otherwise, require a significant degree of coordination among participants, which sparked interest in the study of coordinated online behavior. State-of-the-art methods for detecting coordinated behavior perform static analyses, disregarding the temporal dynamics of coordination. Here, we carry out a dynamic analysis of coordinated behavior. To reach our goal, we build a multiplex temporal network and we perform dynamic community detection to identify groups of users that exhibited coordinated behaviors in time. We find that i) coordinated communities (CCs) feature variable degrees of temporal instability; ii) dynamic analyses are needed to account for such instability, and results of static analyses can be unreliable and scarcely representative of unstable communities; iii) some users exhibit distinct archetypal behaviors that have important practical implications; iv) content and network characteristics contribute to explaining why users leave and join CCs. Our results demonstrate the advantages of dynamic analyses and open up new directions of research on the unfolding of online debates, on the strategies of CCs, and on the patterns of online influence.

@en

In recent years, the frequently reported incidents of Distributed Denial of Service assaults on vehicular networks in various countries have made researchers find new protective solutions. DDoS attacks can propagate through the charging points for electric vehicles in a charging station and affect the production of critical infrastructures such as electric grids. Existing solutions are efficient in attack detection; however, current systems do not offer multi-level protection, and zero-day vulnerabilities are prone to escape from the detection systems. In this paper, we address the problems mentioned above and introduce the first Machine Learning (ML)–based DDoS protective solution to combine prevention and detection mechanisms in vehicular networks. To be more specific, our proposed model is the first to consider the adaptive traffic threshold to generate the alarm for a suspicious amount of traffic flow in an Intrusion Detection Prevention System (IDPS). We call our proposed approach Protecting vEhicular neTworks against distRibuted deniAl of service attacKs (PETRAK). PETRAK uses four functions: prevention, alarm, training, and detection. The alarming system uses the flow parameters and activates the detection module to detect malicious packets. The prevention system works in two modes: immediate and future. PETRAK uses logistic regression to identify incoming packets and signatures of malicious packets to prevent future attacks. We also show that our proposed model is implacable towards advanced post-quantum cryptography-based traffic and also able to analyse side-channel attacks. We run a comprehensive set of experiments to test PETRAK on our synthesized dataset, KDDCUP’99 dataset, CIC-MalMem-2022, and the ToN-IoT dataset. We observe that PETRAK shows an accuracy of 99%. The results claim the efficiency of PETRAK in detecting and preventing DDoS attacks in vehicular networks.

@en

Modeling and predicting the behavior of nodes and users in blockchains provide opportunities for business strategy optimization. Indeed, the number of interactions of a node is strictly related to its balance and its prediction may be used for analytics purposes and investment strategies. However, the amount and diversity of information stored on the blockchain demand advanced tools for the modeling and analysis of blockchain data. Such tools should be able to capture the dynamicity and interaction of multiple independent actors, considering a large number of variables and dynamic interaction graph topologies. This is exacerbated by the use of smart contracts, programs stored in blockchain blocks that bring automation to blockchain's operations and thus increasing the variability of the resulting interaction graphs. Existing modeling methodologies are unable to keep track of all these details, as they are not able to capture the temporal variability of the network. In this paper, we propose a novel framework for modeling and predicting the behavior of smart contracts on a blockchain. We propose the concept of temporal smart contracts networks, i.e., graphs representing the temporal evolution of interactions and data flow. Our framework allows the creation of temporal smart contract networks with different granularity levels by considering different interaction patterns between smart contracts, externally owned accounts, and internal transactions. Thanks to these graphs, we are able to model features such as the node in degree and amount of ether received by a smart contract, which are directly related to its behavior. We incorporate our modeling approach in Ethereum Data Inspection Tool (EDIT), a novel tool able to model interactions and predict them based on historical data. We test different machine learning models to predict features extracted by EDIT, hence allowing for the prediction of the overall behavior of the smart contract. We test EDIT on the Ethereum blockchain and model several temporal smart contracts networks, which represent the interactions and the data flow resulting from about 4 000 000 consecutive blocks. The evaluation of different real case studies shows that the proposed framework is able to predict, with a mean absolute error close to 1%, the evolution of several interesting properties (e.g., amount of received ether) related to both accounts and smart contracts.

@en

Federated Learning (FL) represents the de facto approach for distributed training of machine learning models. Nevertheless, researchers have identified several security and privacy FL issues. Among these, the lack of anonymity exposes FL to linkability attacks, representing a risk for model alteration and worker impersonation, where adversaries can explicitly select the attack target, knowing its identity. Named-Data Networking (NDN) is a novel networking paradigm that decouples the data from its location, anonymising the users. NDN embodies a suitable solution to ensure workers’ privacy in FL, thus fixing the abovementioned issues. However, several issues must be addressed to fit FL logic in NDN semantics, such as missing push-based communication in NDN and anonymous NDN naming convention. To this end, this paper contributes a novel anonymous-by-design FL framework with a customised communication protocol leveraging NDN. The proposed communication scheme encompasses an ad-hoc FL-oriented naming convention and anonymity-driven forwarding and enrollment procedures. The anonymity and privacy requirements considered during the framework definition are fully satisfied through a detailed analysis of the framework's robustness. Moreover, we compare the proposed mechanism and state-of-the-art anonymity solutions, focusing on the communication efficiency perspective. The simulation results show latency and training time improvements up to ∼30%, especially when dealing with large models, numerous federations, and complex networks.

@en

Video surveillance systems provide means to detect the presence of potentially malicious drones in the surroundings of critical infrastructures. In particular, these systems collect images and feed them to a deep-learning classifier able to detect the presence of a drone in the input image. However, current classifiers are not efficient in identifying drones that disguise themselves with the image background, e.g., hiding in front of a tree. Furthermore, video-based detection systems heavily rely on the image’s brightness, where darkness imposes significant challenges in detecting drones. Both these phenomena increase the possibilities for attackers to get close to critical infrastructures without being spotted and hence be able to gather sensitive information or cause physical damages, possibly leading to safety threats. In this article, we propose RANGO, a drone detection arithmetic able to detect drones in challenging images where the target is difficult to distinguish from the background. RANGO is based on a deep learning architecture that exploits a Preconditioning Operation (PREP) that highlights the target by the difference between the target gradient and the background gradient. The idea is to highlight features that will be useful for classification. After PREP, RANGO uses multiple convolution kernels to make the final decision on the presence of the drone. We test RANGO on a drone image dataset composed of multiple already-existing datasets to which we add samples of birds and planes. We then compare RANGO with multiple currently existing approaches to show its superiority. When tested on images with disguising drones, RANGO attains an increase of 6.6% mean Average Precision (mAP) compared to YOLOv5 solution. When tested on the conventional dataset, RANGO improves the mAP by approximately 2.2%, thus confirming its effectiveness also in the general scenario.

@en

Software-defined networking (SDN), enabled by high-performance programmable switches, offers a new avenue to counter cyber attacks. Programmable switches offer the ability to customize and conduct in-depth packet analysis, thus providing efficient and timely responses to DDoS attacks. However, implementing sophisticated DDoS detection may be a challenge in programmable switches because the p4 language does not support floating-point arithmetic, logarithmic functions, or loops. Furthermore, the limited SRAM and TCAM memory on programmable switches makes storing the network connection state difficult. Hence, effective deployment of DDoS detection techniques remains challenging due to these limitations and the rising complexity of the attacks. Many researchers proposed the DDoS detection solution directly on a programmable switch, ignoring the pressing need for a distributed solution. Therefore, this paper presents an innovative, decentralized traffic analysis framework called SPARQ that optimally utilizes the data and control planes. SPARQ is based on Rényi entropy that filters TCP SYN DDoS attacks. It leverages the programming ability of data planes for traffic classification and utilizes the control plane to calculate the metrics and acyclic redundancy checks within the traffic. Moreover, SPARQ uses quartile ranges to track packet inter-arrival time so that abnormal traffic patterns can be identified. We implement SPARQ in a BMv2 switch using the p4runtime controller, enabling seamless integration with SDN systems. We compare the performance of SPARQ with state-of-the-art solutions using the CAIDA dataset. The comparative analysis demonstrates that SPARQ provides a 20.59% reduction in CPU load, an average detection time shorter than 88%, and a 17.8% improvement in true positive rate (TPR).

@en

With the increasing complexity and frequency of cyber attacks, organizations recognize the need for a proactive and targeted approach to safeguard their digital assets and operations. Every industry faces a distinct array of threats shaped by factors such as its industrial objective, geographic footprint, workforce size, revenue, partnerships, and the extent of its digital assets. This results in a wide heterogeneity in threat landscapes, which necessitates tailored threat intelligence sources. While some security practitioners may gravitate towards extensive sources, relying solely on volume-based solutions often leads to “alert fatigue”. For this reason, organization-specific threat intelligence has acquired a growing importance in cybersecurity defense. This work presents a complete and novel framework called OSTIS (Organization-Specific Threat Intelligence System) for generating and managing organization-specific Cyber Threat Intelligence (CTI) data. Our approach identifies reliable security blogs from which we gather CTI data through a custom and focused Web Crawler. Relevant content from such sources is, then, identified and extracted using automated deep-learning models. Moreover, our AI-driven solution maps CTI data to specific domain scenarios, such as education, finance, government, healthcare, industrial control systems, and IoT. To validate and gain insights from the trained models, we also include an explainable AI (XAI, for short) task carried out by leveraging the SHapley Additive exPlanations (SHAP) tool. This allows us to interpret the prediction process and discern influential content from data. The last step of our framework consists of the generation of an Organization Specific Threat Intelligence Knowledge Graph (OSTIKG), empowering organizations to identify and visualize attack patterns and incidents, promptly. To create this graph, we develop and adapt several techniques to extract diverse entities, including malware groups, campaigns, attack types, malware types, software tools, and so forth, and to identify relationships among them. Finally, through an extensive experimental campaign, we certify the validity and performance of all the components of our framework, which shows a 0.84 F1-score in the identification of relevant content, a 0.93 F1-score for the domain classification, and a 0.95 and 0.89 F1-score in the identification of entities and relations to build our OSTIKG graph.

@en

In the past decades, the rise of artificial intelligence has given us the capabilities to solve the most challenging problems in our day-to-day lives, such as cancer prediction and autonomous navigation. However, these applications might not be reliable if not secured against adversarial attacks. In addition, recent works demonstrated that some adversarial examples are transferable across different models. Therefore, it is crucial to avoid such transferability via robust models that resist adversarial manipulations. In this paper, we propose a feature randomization-based approach that resists eight adversarial attacks targeting deep learning models in the testing phase. Our novel approach consists of changing the training strategy in the target network classifier and selecting random feature samples. We consider the attacker with a Limited-Knowledge and Semi-Knowledge conditions to undertake the most prevalent types of adversarial attacks. We evaluate the robustness of our approach using the well-known UNSW-NB15 datasets that include realistic and synthetic attacks. Afterward, we demonstrate that our strategy outperforms the existing state-of-the-art approach, such as the Most Powerful Attack, which consists of fine-tuning the network model against specific adversarial attacks. Further, we demonstrate the practicality of our approach using the VIPPrint dataset through a comprehensive set of experiments. Finally, our experimental results show that our methodology can secure the target network and resists adversarial attack transferability by over 60%.

@en

We present a device-centric analysis of security and privacy attacks and defenses on extended reality (XR) devices. We present future research directions and propose design considerations to help ensure the security and privacy of XR devices.

@en

Dynamic Wireless Power Transfer (DWPT) is a novel technology that allows charging an electric vehicle while driving thanks to a dedicated road infrastructure. DWPT's capabilities in automatically establishing charging sessions and billing without users' intervention make it prone to cybersecurity attacks. Hence, security is essential in preventing fraud, impersonation, and user tracking. To this aim, researchers proposed different solutions for authenticating users. However, recent advancements in quantum computing jeopardize classical public key cryptography, making currently existing solutions in DWPT authentication nonviable. To avoid the resource burden imposed by technology upgrades, it is essential to develop post-quantum-resistant solutions. In this paper, we propose DynamiQS, the first post-quantum secure authentication protocol for dynamic wireless charging. DynamiQS is privacy-preserving and secure against attacks on the DWPT. We leverage an Identity-Based Encryption with Lattices in the Ring Learning With Error framework. Furthermore, we show the possibility of using DynamiQS in a real environment, leveraging the results of cryptographic computation on real constrained devices and simulations. DynamiQS reaches a total time cost of around 281 ms, which is practicable in dynamic charging settings (car and charging infrastructure).@en

In the age of digital music streaming, playlists on platforms like Spotify have become an integral part of individuals’ musical experiences. People create and publicly share their own playlists to express their musical tastes, promote the discovery of their favorite artists, and foster social connections. In this work, we aim to address the question: can we infer users’ private attributes from their public Spotify playlists? To this end, we conducted an online survey involving 739 Spotify users, resulting in a dataset of 10,286 publicly shared playlists comprising over 200,000 unique songs and 55,000 artists. Then, we utilize statistical analyses and machine learning algorithms to build accurate predictive models for users’ attributes.

@en

Threshold signature is a powerful cryptographic technique with a large number of real-life applications. As designed by Boneh and Komlo (CRYPTO’22), TAPS is a new threshold signature integrating privacy and accountability. It allows a combiner to combine t signature shares while protecting t and the signing group from the public. It also enables a tracer to trace a threshold signature to its original signing group. Despite being valuable, TAPS neglects the witnessing of tracing, i.e., leaves the tracing activity unrestrained. In this paper, we introduce Accountable and Private Threshold Signature with Hidden Witnesses (HiTAPS) that not only provides privacy and accountability, but also incorporates witnessed tracing. In specific, we first utilize Dynamic Threshold Public-Key Encryption (DTPKE) and ElGamal encryption to designate a set of t^′ witnesses for endorsing the tracing activity. We then compute a keyed-hash tag for the t^′ witnesses to initiate the tracing activity secretly. Moreover, we present an optimized protocol HiTAPS2 to reduce communication overhead of the combiner. We formalize the definitions, security, and privacy for HiTAPS. We formally prove its security and privacy. To evaluate the performance of HiTAPS and HiTAPS2, we build a prototype based on pypbc. Experimental results show that HiTAPS takes 217(370) ms to combine (track) a threshold signature of 5 signers (witnesses). The optimized HiTAPS2 only takes 137 ms to combine a threshold signature of 5 signers.

@en

Private set intersection protocols allow two parties with private sets of data to compute the intersection between them without leaking other information about their sets. These protocols have been studied for almost 20 years, and have been significantly improved over time, reducing both their computation and communication costs. However, when more than two parties want to compute a private set intersection, these protocols are no longer applicable. While extensions exist to the multi-party case, these protocols are significantly less efficient than the two-party case. It remains an open question to design collusion-resistant multi-party private set intersection (MPSI) protocols that come close to the efficiency of two-party protocols. This work is made more difficult by the immense variety in the proposed schemes and the lack of systematization. Moreover, each new work only considers a small subset of previously proposed protocols, leaving out important developments from older works. Finally, MPSI protocols rely on many possible constructions and building blocks that have not been summarized. This work aims to point protocol designers to gaps in research and promising directions, pointing out common security flaws and sketching a frame of reference. To this end, we focus on the semi-honest model. We conclude that current MPSI protocols are not a one-size-fits-all solution, and instead there exist many protocols that each prevail in their own application setting.

@en

Existing literature on adversarial Machine Learning (ML) focuses either on showing attacks that break every ML model or defenses that withstand most attacks. Unfortunately, little consideration is given to the actual feasibility of the attack or the defense. Moreover, adversarial samples are often crafted in the “feature-space,” making the corresponding evaluations of questionable value. Simply put, the current situation does not allow one to estimate the actual threat posed by adversarial attacks, leading to a lack of secure ML systems.

We aim to clarify such confusion in this article. By considering the application of ML for Phishing Website Detection (PWD), we formalize the “evasion-space,” in which an adversarial perturbation can be introduced to fool an ML-PWD—demonstrating that even perturbations in the “feature-space” are useful. Then, we propose a realistic threat model describing evasion attacks against ML-PWD that are cheap to stage, and hence intrinsically more attractive for real phishers. After that, we perform the first statistically validated assessment of state-of-the-art ML-PWD against 12 evasion attacks. Our evaluation shows (i) the true efficacy of evasion attempts that are more likely to occur; and (ii) the impact of perturbations crafted in different evasion-spaces; our realistic evasion attempts induce a statistically significant degradation (3–10% at p< 0.05), and their cheap cost makes them a subtle threat. Notably, however, some ML-PWD are immune to our most realistic attacks (p = 0.22).

Finally, as an additional contribution of this journal publication, we are the first to propose and empirically evaluate the intriguing case wherein an attacker introduces perturbations in multiple evasion-spaces at the same time. These new results show that simultaneously applying perturbations in the problem- and feature-space can cause a drop in the detection rate from 0.95 to 0.

Our contribution paves the way for a much-needed re-assessment of adversarial attacks against ML systems for cybersecurity.@en

The Internet of Things rapid growth poses privacy and security challenges for the traditional key storage methods. Physical Unclonable Functions offer a potential solution but require secure fuzzy extractors to ensure reliable replication. This paper introduces X-Lock, a novel and secure computational fuzzy extractor that addresses the limitations faced by traditional solutions in resource-constrained IoT devices. X-Lock offers a reusable and robust solution, effectively mitigating the impacts of bias and correlation through its design. Leveraging the preferred state of a noisy source, X-Lock encrypts a random string of bits that can be later used as seed to generate multiple secret keys. To prove our claims, we provide a comprehensive theoretical analysis, addressing security considerations, and implement the proposed model. To evaluate the effectiveness and superiority of our proposal, we also provide practical experiments and compare the results with existing approaches. The experimental findings demonstrate the efficacy of our algorithm, showing comparable memory cost (≈2.4 KB for storing 5 keys of 128 bits) while being 3 orders of magnitude faster with respect to the state-of-the-art solution (0.086 ms against 15.51 s).

@en

Machine learning based phishing website detectors (ML-PWD) are a critical part of today's anti-phishing solutions in operation. Unfortunately, ML-PWD are prone to adversarial evasions, evidenced by both academic studies and analyses of real-world adversarial phishing webpages. However, existing works mostly focused on assessing adversarial phishing webpages against ML-PWD, while neglecting a crucial aspect: investigating whether they can deceive the actual target of phishing - -the end users. In this paper, we fill this gap by conducting two user studies (n=470) to examine how human users perceive adversarial phishing webpages, spanning both synthetically crafted ones (which we create by evading a state-of-the-art ML-PWD) as well as real adversarial webpages (taken from the wild Web) that bypassed a production-grade ML-PWD. Our findings confirm that adversarial phishing is a threat to both users and ML-PWD, since most adversarial phishing webpages have comparable effectiveness on users w.r.t. unperturbed ones. However, not all adversarial perturbations are equally effective. For example, those with added typos are significantly more noticeable to users, who tend to overlook perturbations of higher visual magnitude (such as replacing the background). We also show that users' self-reported frequency of visiting a brand's website has a statistically negative correlation with their phishing detection accuracy, which is likely caused by overconfidence. We release our resources.

@en