The increasing integration of digital technologies in connected vehicles introduces cybersecurity risks that extend beyond individual vehicles, with the potential to disrupt entire transportation systems. Current practice (e.g., ISO/SAE 21434 TARA) focuses on threat identification and qualitative impact ratings at the vehicle boundary, with limited systemic quantification. This study presents a systematic, simulation-based methodology for quantifying the systemic operational and safety impacts of cyber threats on connected vehicles, evaluating cascading effects across the transport network. Three representative scenarios are examined: (I) telematics-induced sudden braking causing a cascading collision, (II) remote disabling on a motorway (M25) segment, and (III) a compromised Roadside Unit (RSU) spoofing Variable Speed Limit (VSL) and phantom lane closure messages to connected and automated vehicles (CAVs). The results highlight the potential for cascading safety incidents and systemic operational degradation, as evidenced by the defined systemic operational and safety vectors, factors that are insufficiently addressed in the current scope of the ISO/SAE 21434 standard, which primarily focuses on individual vehicle-level threats. The findings underscore the need to incorporate systemic evaluation into existing frameworks to enhance cyber resilience across connected vehicle ecosystems. The framework complements ISO/SAE 21434 by supplying quantitative, reproducible evidence for the impact rating step at a systemic scale, reducing assessor subjectivity and supporting policy and operations, enabling more data-driven evaluations of systemic cyber risks.

The increasing complexity and connectivity of modern vehicles have made automotive networks, particularly the Controller Area Network (CAN) bus, vulnerable to cyberattacks. Fuzzing is a critical technique for proactively finding security weaknesses, but traditional methods are inefficient and struggle to scale with the complexity of modern vehicles. This paper introduces AMCFF-RL, an adaptive framework that uses Deep Reinforcement Learning (DRL) with multi-modal feature extraction to systematically analyse for vulnerabilities. Rather than relying on unguided or purely random fuzzing, AMCFF-RL integrates multi-modal feature extraction with DRL and advanced visualization, allowing it to learn and adapt its strategy based on real-time feedback from the network and thereby improve the efficiency and effectiveness of the fuzzing process. Comprehensive visualization tools serve a dual purpose: they offer human-interpretable insights while also generating rich feature representations that support the anomaly detection pipeline and the DRL agent.

Multi-agent deep reinforcement learning (MARL) extends deep reinforcement learning (DRL) to environments involving multiple interacting agents and has enabled applications in domains such as autonomous vehicles, robotics, unmanned aerial vehicles (UAVs), and multi-player games. Compared with single-agent learning, MARL introduces additional challenges, including non-stationarity, partial observability, multi-agent credit assignment, and scalability. This paper presents a narrative survey of recent developments in MARL and discusses major approaches proposed to address these challenges. In particular, we examine research directions centred on centralised training with decentralised execution (CTDE), value decomposition, learned communication, graph-based methods, and model-based learning. We further discuss commonly used benchmark environments and evaluation practices, highlighting considerations related to reproducibility, robustness, and generalisation. Finally, we outline open research challenges and future directions concerning theoretical understanding, sample efficiency, scalable coordination, and deployment in real-world settings. Rather than providing an exhaustive systematic review, this survey aims to offer an organised and up-to-date synthesis of recent progress in MARL.

The use of ontologies in context-aware systems has been widely recognised across various applications and domains. However, their potential for enhancing cybersecurity in the automotive domain remains underexplored. Understanding how semantic knowledge can be acquired, structured, and leveraged to support security-related decision-making is a critical area of research. While artificial intelligence, and machine learning-based approaches are commonly employed, the advantages of semantic knowledge-based models, particularly ontologies, require attention and practical implementation. This paper emphasises the significance of utilising context-aware ontologies for automotive cybersecurity. Specifically, it explores the development of ontology-based models enriched with contextual information and demonstrates their potential to enhance automotive security measures.

As connected vehicles (CVs) play an increasingly pivotal role in modern transportation, cybersecurity threats targeting these systems have become a critical area of concern. This study systematically identifies and classifies vulnerabilities from the National Vulnerability Database (NVD) and the Automotive Attack Database (AAD) using a semi-automated filtering process. Our analysis identifies a total of 508 vulnerabilities across these databases, which are categorised based on ISO/SAE 21434 impact categories: safety, financial, operational, and privacy. A key finding reveals that 14.6

Parking availability prediction is a critical component of intelligent transportation systems, aiming to reduce congestion and improve urban mobility. While traditional deep learning models such as Long Short-Term Memory (LSTM) networks have been widely applied, they lack mechanisms to quantify uncertainty, limiting their robustness in real-world deployments. This paper proposes a Bayesian Neural Network (BNN)-based framework for parking occupancy prediction that explicitly models both epistemic and aleatoric uncertainty. Although BNNs have shown promise in other domains, they remain underutilised in parking prediction—likely due to the computational complexity and the absence of real-time context integration in earlier approaches. Our approach leverages contextual features, including temporal and environmental factors, to enhance uncertainty-aware predictions. The framework is evaluated under varying data conditions, including data scarcity (90%, 50%, and 10% of training data) and synthetic noise injection to simulate aleatoric uncertainty. Results demonstrate that BNNs outperform other methods, achieving an average accuracy improvement of 27.4% in baseline conditions, with consistent gains under limited and noisy data. Applying uncertainty thresholds at 20% and 30% further improves reliability by enabling selective, confidence-based decision making. This research shows that modelling both types of uncertainty leads to significantly improved predictive performance in intelligent transportation systems and highlights the potential of uncertainty-aware approaches as a foundation for future work on integrating BNNs with hybrid neuro-symbolic reasoning to enhance decision making under uncertainty.

As our dependence on the internet and digital platforms grows, the risk of cyber threats rises, making it essential to implement effective measures to safeguard sensitive information through cybersecurity, ensure system integrity, and prevent unauthorized data access. Fuzz testing, commonly known as fuzzing, is a valuable technique for software testing as it uncovers vulnerabilities and defects in systems by introducing random data inputs, often leading to system crashes. In the Internet of Things (IoT) domain, fuzzing is crucial for identifying vulnerabilities in networks, devices, and applications through automated tools that systematically inject malformed inputs into IoT systems. However, despite its importance, existing research on fuzzing techniques in IoT contexts remains limited by the absence of standardized benchmarks, inefficiencies in re-hosting strategies, and difficulties in detecting complex, condition-dependent vulnerabilities. The primary objective of this study is to comprehensively evaluate current fuzzing practices, emphasizing adaptive techniques designed for IoT systems. Using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) model, a systematic literature review was conducted across 32 academic articles published between 2020 and 2024. The analysis revealed that although fuzzing enhances IoT security, its effectiveness is hindered by device heterogeneity, limited system resources, and evolving cyber threat landscapes. The findings suggest that to overcome these limitations, future research should focus on AI-driven fuzzing methods, robust multi-architecture support, and the development of standardized evaluation frameworks to strengthen IoT cybersecurity.

Combinatorial optimization problems are inherently complex and difficult to solve. Although evolutionary and swarm intelligence algorithms have proven effectiveness, achieving an optimal balance between exploration and exploitation (EvE) remains a significant challenge. The growing trend of employing multiple operators with adaptive operator selection schemes aims to address EvE issues; however, there remains a significant demand for utilizing tailored adaptive selection schemes in search. Reinforcement Learning (RL) has recently been proposed as an efficient technique to customize and develop highly effective and adaptive selection mechanisms. However, scalability continuous to pose a significant challenge. This paper presents a novel RL-based approach that establishes a generalized and scalable framework for acquiring, processing, and leveraging experiential knowledge for both immediate and long-term use. The experimental results provide evidence of a certain level of success in validating the proposed framework.

Predicting the future location of mobile objects reinforces location-aware services with proactive intelligence and helps businesses and decision-makers with better planning and near real-time scheduling in different applications such as traffic congestion control, location-aware advertisements and monitoring public health and well-being. Recent developments in smartphone and location sensors technology and the prevalence of using location-based social networks alongside the improvements in AI and machine learning techniques provide an excellent opportunity to exploit massive amounts of historical and real-time contextual information to recognise mobility patterns and achieve more accurate and intelligent predictions. This unique survey provides a comprehensive overview of the next useful location prediction problem with context-awareness and the related studies. First, we explain the concepts of context and context-awareness and define the next location prediction problem. Then we analyse more than 30 studies in this field concerning the prediction method, the challenges addressed, the datasets and metrics used for training and evaluating the model and the types of context incorporated. Finally, we discuss the advantages and disadvantages of different approaches, focusing on the usefulness of the predicted location and identifying the open challenges and future work on this subject.

The growing complexity of vehicle network connectivity has broadened the cyber-threat landscape, introducing substantial safety risks for both passengers and the environment. Traditional security mechanisms, relying on rigid decision-making processes, often fail to address the demands of this dynamic and interconnected ecosystem. To effectively manage emerging security threats and adapt to diverse scenarios, integrating context awareness has become crucial. Context-aware systems can typically adapt their behaviour in response to changes in their surrounding environment using context information. Ontologies serve as powerful tools for modelling and reasoning of context information. However, existing ontology-based context-aware security models are constrained by static thresholds and fail to adapt the rapid changes in real-time. This paper introduces a dynamic context-aware real-time security model for the automotive domain. By leveraging a Python-based implementation alongside OWL 2 RL Ontology model, the proposed approach dynamically adapts context information based on live data for security analysis. The applicability and effectiveness of the proposed approach is demonstrated using a use case of EV charging process.

Vehicular systems are becoming interconnected digital ecosystems, reliant on complex networks for operations and safety. Cybersecurity vulnerabilities in these networks can jeopardize road safety. This study introduces a fuzzing framework to identify vulnerabilities in the Controller Area Network (CAN) bus. Using Automated Reverse Engineering-Guided Fuzzing, modified data packets are injected into the CAN framework, and Electronic Control Units (ECUs) reactions are monitored to uncover vulnerabilities. This approach identifies CAN bus network weaknesses and enhances understanding of their operational characteristics, setting a new standard for automotive cybersecurity.

The automotive industry increasingly relies on Over-the-Air (OTA) updates to deliver essential security patches to vehicles. However, this dependence may introduce significant cy-bersecurity vulnerabilities, particularly concerning the integrity and privacy of updates. This paper presents an integrated framework that combines threat modeling and formal verification by employing an identical system model across all stages. The process begins with applying the ThreatGet tool to identify potential threats in the OTA update process, which directly guide the formulation of formal security requirements expressed as Computation Tree Logic (CTL) properties. The same high-level model encompassing a Cloud Server (CS), Telematics Control Unit (TCU), Central Gateway Unit (CGU), Advanced Driver Assistance System (ADAS) module, and an attacker module is consistently used for both threat modeling and encoding in the NuSMV model checker. This unified approach ensures that identified threats translate seamlessly into verifiable properties. Experimental threat analysis and verification results demonstrate the effectiveness of our integrated approach in uncovering OTA update vulnerabilities properties.

The rise in Connected and Automated Vehicles (CAVs) and Intelligent Transport Systems (ITSs) introduced by OEMs has increased the demand for modern vehicle sophistication. This sophistication involves a variety of software capabilities and functionalities embedded in over 100 ECUs in a vehicle. This has led to the need for over-the-air (OTA) updates. OTA updates can be delivered wirelessly, eliminating the need to bring vehicles to the garage for updates. This is more convenient for owners, reduces costs for OEMs, and reduces greenhouse gas emissions. There exist different OTA update considerations that are adopted by automotive OEMs, such as the Uptane framework, Open Mobile Alliance Device Management (OMA-DM) standard, and the general ISO 24089 standard, including subvariance of Uptane and OMA-DM. However, the systematic implementation of security-by-design applying ISO 21434 in OTA systems is less employed, and there remains a gap in this practice of security-by-design that the automotive industry can adapt to ensure a systematic approach to secure OTA update technology. OTA update security hinges on identifying vulnerability pathways for potential malicious attacks. Therefore, identifying and mitigating potential vulnerabilities throughout the OTA update process is critical for robust security. This paper proposes an adaptable security-by-design approach to OTA update, built and extended from our work Iyieke et al. (2023). The adaptable security-by-design approach is then applied to a developed prototype OTA update system based on the Uptane framework as implemented by Toradex. Security-by-design is a well-established concept in enterprise systems, but is still developing in the cyber–physical system of automotive cybersecurity. Our proposed approach covers the security engineering lifecycle, the logical security layered concept, and the security architecture. A threat analysis and risk assessment (TARA) is performed based on the international automotive cybersecurity standard ISO/SAE 21434. The highest threats identified from the TARA are formalized, and corresponding mitigation actions are defined according to UNECE WP29. Penetration testing is conducted to verify the approach’s capability to reinforce the security of the OTA update systems against some of the identified risks and threats. Our proposed approach provides a systematic and adaptable security-by-design approach to ensure secure OTA updates in modern vehicles; OEMs and other stakeholders can use it to develop secure OTA systems regardless of the OTA update technology used.

The rapid advancement and deployment of connected and autonomous vehicles (CAVs) present transformative opportunities to enhance safety, efficiency, and convenience within the transportation industry. However, these innovations introduce significant cybersecurity risks due to the complex electronics and continuous connectivity that CAVs depend on. Traditional testing methods, while critical, often fall short in detecting vulnerabilities across the vast range of scenarios these vehicles may encounter. Formal verification, a mathematical approach to system validation, offers a more rigorous and comprehensive solution by ensuring that systems operate as expected to search through all possible execution paths. However, defining appropriate system properties for verification remains a challenge, as a system designer may write properties that fail to address real-world threats effectively. This research addresses this gap by integrating threat analysis into the process of defining security properties, ensuring that the verification process is aligned with actual cybersecurity risks. We leverage Natural Language Processing (NLP) to extract key security details from threat analysis result texts, automating the generation of system properties. This approach simplifies the verification process, with its usability demonstrated through a high-level 5G-V2X design use case scenario.

The Road Vehicles Cybersecurity Engineering Standard, ISO/SAE 21434, provides a framework for road vehicle Threat Analysis and Risk Assessment (TARA). The TARA framework must include Connected Vehicles (CVs) and their connectivity with external interfaces. However, assessing cyber-attack feasibility on CVs is a significant challenge, as traditionally, qualitative and subjective expert opinions are the norm. Additionally, there is a need for historical data on security-related incidents and dynamically evolving interconnected vehicle-to-everything (V2X) entities for feasibility assessment, which is not readily available. To address this problem, this paper presents, to the best of our knowledge, the first simulation-based TARA framework designed to characterise, quantify, and assess the Window of Opportunity (WO) for attackers—a metric that indicates the likelihood of an attack. A case study involving Bluetooth, with one attacker and one target, is modelled to demonstrate the proposed framework WOLVES’s applicability. Two scenarios have been investigated using different motorway roads in the UK. The primary outcome is the WOLVES framework, which employs a data-driven approach using both prior and likelihood information to estimate the probability of a successful cyber attack on a given technology in CVs. The findings from this research could assist threat analysts, decision-makers, and planners involved in CV risk assessment by enhancing the modelling of attack feasibility for cybersecurity threats in dynamic scenarios and developing appropriate mitigation strategies.

Accurate parking availability prediction is critical for intelligent transportation systems, but real-world deployments often face data sparsity, noise, and unpredictable changes. Addressing these challenges requires models that are not only accurate but also uncertainty-aware. In this work, we propose a loosely coupled neuro-symbolic framework that integrates Bayesian Neural Networks (BNNs) with symbolic reasoning to enhance robustness in uncertain environments. BNNs quantify predictive uncertainty, while symbolic knowledge-extracted via decision trees and encoded using probabilistic logic program-ming-is leveraged in two hybrid strategies: (1) using symbolic reasoning as a fallback when BNN confidence is low, and (2) refining output classes based on symbolic constraints before reapplying the BNN. We evaluate both strategies on real-world parking data under full, sparse, and noisy conditions. Results demonstrate that both hybrid methods outperform symbolic reasoning alone, and the context-refinement strategy consistently exceeds the performance of Long Short-Term Memory (LSTM) networks and BNN baselines across all prediction windows. Our findings highlight the potential of modular neuro-symbolic integration in real-world, uncertainty-prone prediction tasks.

This demonstration introduces an adaptive fuzzing physical test bed aimed at identifying vulnerabilities within automotive systems, specifically focusing on the Controller Area Network (CAN) bus. By employing "Automated Reverse Engineering-Guided Fuzzing" (ARE-GF), our framework evaluates the security resilience of the CAN network against sophisticated attacks. The demo showcases live demonstrations of the fuzzing process, the creation of the test bed using cost-effective electrical components, real-time ECU response analysis, and examples of discovered vulnerabilities, providing insights into advanced automotive cybersecurity testing methodologies.

Vehicular Ad-Hoc Networks (VANETs), a subset of Mobile Ad-Hoc Networks (MANETs), are wire- less networks formed around moving vehicles, enabling communication between vehicles, roadside infrastructure, and servers. With the rise of autonomous and connected vehicles, security concerns surrounding VANETs have grown. VANETs still face challenges related to privacy with full-scale deployment due to a lack of user trust. Critical factors shaping VANETs include their dynamic topology and high mobility characteristics. Authentication protocols emerge as the cornerstone of enabling the secure transmission of entities within a VANET. Despite concerted efforts, there remains a need to incorporate verification approaches for refining authentication protocols. Formal verification constitutes a mathematical approach enabling developers to validate protocols and rectify design errors with precision. Therefore, this review focuses on authentication protocols as a pivotal element for securing entity transmission within VANETs. It presents a comparative analysis of existing protocols, identifies research gaps, and introduces a novel framework that incorporates formal verification and threat modeling. The review considers key factors influencing security, sheds light on ongoing challenges, and emphasises the significance of user trust. The proposed framework not only enhances VANET security but also contributes to the growing field of formal verification in the automotive domain. As the outcomes of this study, several research gaps, challenges, and future research directions are identified. These insights would offer valuable guidance for researchers to establish secure authentication communication within VANETs.

Recent advancements in technology have transformed conventional mechanical vehicles into sophisticated computer systems on wheels. This transition has elevated their intelligence and facilitated seamless connectivity. However, such development has also escalated the possibility of compromising the vehicle’s cyber security expanding the overall cyber threat landscape. This necessitates an increased demand for security measures that manifest flexibility and adaptability instead of static threshold-based measures. Context-awareness techniques can provide a promising direction for such security solutions. Integration of context-awareness in security analysis helps in analysing the behaviour of the environment where IoT devices are deployed, enabling adaptive decision-making that aligns with the current situation. While the incorporation of context-awareness into adaptive systems has been explored extensively, its application to support the cyber security of vehicular ecosystem is relatively new. In this paper, we proposed a context-aware conceptual framework for automotive vehicle security that allows us to analyse real-time situations thereby identifying security threats. The usability of the framework is demonstrated considering an Electric Vehicle(EV) Charging case study.

Combinatorial optimisation problems are known as unpredictable and challenging due to their nature and complexity. One way to reduce the unpredictability of such problems is to identify features and the characteristics that can be utilised to guide the search using domain-knowledge and act accordingly. Many problem solving algorithms use multiple complementary operators in patterns to handle such unpredictable cases. A well-characterised search space may help to evaluate the problem states better and select/apply a neighbourhood operator to generate more productive new problem states that allow for a smoother path to the final/optimum solutions. This applies to the algorithms that use multiple operators to solve problems. However, the remaining challenge is determining how to select an operator in an optimal way from the set of operators while taking the search space conditions into consideration. Recent research shows the success of adaptive operator selection to address this problem. However, efficiency and scalability issues persist in this regard. In addition, selecting the most representative features remains crucial in addressing problem complexity and inducing commonality for transferring experience across domains. This paper investigates if a problem can be represented by a number of features identified by landscape analysis, and whether an adaptive operator selection scheme can be constructed using Machine Learning (ML) techniques to address the efficiency and scalability problem. The proposed method determines the optimal categorisation by analysing the predictivity of a set of features using the most well-known supervised ML techniques. The identified set of features is then used to construct an adaptive operator selection scheme. The findings of the experiments demonstrate that supervised ML algorithms are highly effective when building adaptable operator selectors.