Vasileios Kouliaridis

The evolution of cybersecurity is closely linked to the development and improvement of artificial intelligence (AI). As a key tool for realizing more cybersecure ecosystems, Intrusion Detection Systems (IDSs) have evolved tremendously in recent years by integrating machine learning (ML) techniques to detect increasingly sophisticated cybersecurity attacks hidden in big data. However, traditional approaches rely on centralized learning, in which data from end nodes are shared with data centers for analysis. Recently, the application of federated learning (FL) in this context has attracted great interest to come up with collaborative intrusion detection approaches where data does not need to be shared. Due to the recent rise of this field, this work presents a complete, contemporary taxonomy for FL-enabled IDS approaches that stems from a comprehensive survey of the literature from 2018 to 2022. Precisely, our discussion includes an analysis of the main ML models, datasets, aggregation functions, as well as implementation libraries employed by the proposed FL-enabled IDS approaches. On top of everything else, we provide a critical view of the current state of the research around this topic, and describe the main challenges and future directions based on the analysis of the literature and our own experience in this area.

The increasing frequency of attacks on Android applications coupled with the recent popularity of large language models (LLMs) necessitates a comprehensive understanding of the capabilities of the latter in identifying potential vulnerabilities, which is key to mitigate the overall risk. To this end, the work at hand compares the ability of nine state-of-the-art LLMs to detect Android code vulnerabilities listed in the latest Open Worldwide Application Security Project (OWASP) Mobile Top 10. Each LLM was evaluated against an open dataset of over 100 vulnerable code samples, assessing each model’s ability to identify key vulnerabilities. Our analysis reveals the strengths and weaknesses of each LLM, identifying important factors that contribute to their performance. Additionally, we offer insights into context augmentation with retrieval-augmented generation (RAG) for detecting Android code vulnerabilities, which in turn may propel secure application development. Finally, while the reported findings regarding code vulnerability analysis show promise, they also reveal significant discrepancies among the different LLMs.

With the increasing use of smartphones for a wide variety of online services, states and countries are issuing official applications to store government-issued documents that can be used for identification (e.g., electronic identity cards), health (e.g., vaccination certificates), and transport (e.g., driver’s licenses). However, the privacy and security risks associated with the storage of sensitive personal information on such apps are a major concern. This work presents a thorough analysis of official Android wallet apps, focusing mainly on apps used to store identification documents and/or driver’s licenses. Specifically, we examine the security and privacy level of such apps using three analysis tools and discuss the key findings and the risks involved. We additionally explore Android app security best practices and various security measures that can be employed to mitigate these risks, such as updating deprecated components and libraries. Altogether, our findings demonstrate that, while there are various security measures available, there is still a need for more comprehensive solutions to address the privacy and security risks associated with the use of Android wallet apps.

Nowadays, with the increasing usage of Android devices in daily life activities, malware has been increasing rapidly, putting peoples’ security and privacy at risk. To mitigate this threat, several researchers have proposed different methods to detect Android malware. Recently, machine learning based models have been explored by a significant mass of researchers checking for Android malware. However, selecting the most appropriate model is not straightforward, since there are several aspects that must be considered. Contributing to this domain, the current paper explores Android malware detection from diverse perspectives; this is achieved by optimizing and evaluating various machine learning algorithms. Specifically, we conducted an experiment for training, optimizing, and evaluating 27 machine learning algorithms, and a Deep Neural Network (DNN). During the optimization phase, we performed hyperparameter analysis using the Optuna framework. The evaluation phase includes the measurement of different performance metrics against a contemporary, rich dataset, to conclude with the most accurate model. The best model was further interpreted by conducting feature analysis, using the Shapley Additive Explanations (SHAP) framework. Our experiment results showed that the best model is the DNN consisting of four layers (two hidden), using the Adamax optimizer, as well as the Binary Cross-Entropy (loss), and the Softsign activation functions. The model succeeded with 86% prediction accuracy, while the balanced accuracy, the F1-score, and the ROC-AUC metrics were at 82%.

Built on top of UDP, the recently standardized QUIC protocol primarily aims to gradually replace the TCP plus TLS plus HTTP/2 model. For instance, HTTP/3 is designed to exploit QUIC’s features, including reduced connection establishment time, multiplexing without head of line blocking, always-encrypted end-to-end security, and others. This work serves two key objectives. Initially, it offers the first to our knowledge full-fledged review on QUIC security as seen through the lens of the relevant literature so far. Second and more importantly, through extensive fuzz testing, we conduct a hands-on security evaluation against the six most popular QUIC-enabled production-grade servers. This assessment identified several effective and practical zero-day vulnerabilities, which, if exploited, can quickly overwhelm the server resources. This finding is a clear indication that the fragmented production-level implementations of this contemporary protocol are not yet mature enough. Overall, the work at hand provides the first wholemeal appraisal of QUIC security from both a literature review and empirical standpoint, and it is therefore foreseen to serve as a reference for future research in this timely area.

Breaches in the cyberspace due to cyber-physical attacks can harm the physical space, and any type of vehicle is an alluring target for wrongdoers for an assortment of reasons. Especially, as the automobiles are becoming increasingly interconnected within the Cooperative Intelligent Transport System (C-ITS) realm and their level of automation elevates, the risk for cyberattacks augments along with the attack surface, thus inexorably rendering the risk of complacency and inaction sizable. Next to other defensive measures, intrusion detection systems (IDS) already comprise an inextricable component of modern automobiles in charge of detecting intrusions in the system while in operation. This work concentrates on in-vehicle IDS with the goal to deliver a fourfold comprehensive survey of surveys on this topic. First, we collect and analyze all existing in-vehicle IDS classifications and fuse them into a simpler, overarching one that can be used as a base for classifying any work in this area. Second, we gather and elaborate on the so-far available datasets which can be possibly used to train and evaluate an in-vehicle IDS. Third, we survey non-commercial simulators which may be utilized for creating a dataset or evaluating an IDS. The last contribution pertains to a thorough exposition of the future trends and challenges in this area. To our knowledge, this work provides the first wholemeal survey on in-vehicle IDS, and it is therefore anticipated to serve as a groundwork and point of reference for multiple stakeholders at varying levels.

Digital COVID-19 certificates serve as reliable proof that an individual was vaccinated, tested negative, or healed from COVID-19, facilitating health, occupational, educational, and travel activities during the pandemic. This paper contributes the first to our knowledge state-of-the-art and holistic review of this ecosystem, attempting to answer the following questions: (i) is there a harmonization among academia, organizations, and governments in terms of the certificate deployment technology?, (ii) what is the proliferation of such schemes worldwide and how similar are they?, and (iii) are smartphone applications that accompany such schemes privacy-preserving from an end-user’s perspective? To respond to these questions, a four-tier approach is followed: (a) we scrutinize the so far academic works suggesting some type of digital certificate, highlighting common characteristics and weaknesses; (b) we constructively report on the different initiatives proposed by organizations or alliances; (c) we briefly review 54 country initiatives around the globe; and (d) we analyze both statically and dynamically all official Android smartphone applications offered for such certificates to reveal possible hiccups affecting the security or privacy of the end-user. From a bird’s eye view, the great majority of the proposed or developed schemes follow either the blockchain model or the asymmetric cryptosystem, the spread of schemes especially in Europe and partly in Asia is high, some degree of distinctiveness among the relevant schemes developed by countries does exist, and there are substantial variations regarding the privacy level of the applications between Europe on the one hand and Asia and America on the other.

Year after year, mobile malware attacks grow in both sophistication and diffusion. As the open source Android platform continues to dominate the market, malware writers consider it as their preferred target. Almost strictly, state-of-the-art mobile malware detection solutions in the literature capitalize on machine learning to detect pieces of malware. Nevertheless, our findings clearly indicate that the majority of existing works utilize different metrics and models and employ diverse datasets and classification features stemming from disparate analysis techniques, i.e., static, dynamic, or hybrid. This complicates the cross-comparison of the various proposed detection schemes and may also raise doubts about the derived results. To address this problem, spanning a period of the last seven years, this work attempts to schematize the so far ML-powered malware detection approaches and techniques by organizing them under four axes, namely, the age of the selected dataset, the analysis type used, the employed ML techniques, and the chosen performance metrics. Moreover, based on these axes, we introduce a converging scheme which can guide future Android malware detection techniques and provide a solid baseline to machine learning practices in this field.