Telecommunication networks are highly vulnerable to disasters, disrupting critical services and causing significant economic losses. This paper presents a constraintbased Integer Linear Programming (ILP) model for disaster recovery that jointly optimizes risk-aware routing and revenue maximization under bandwidth constraints. Unlike traditional heuristic approaches, our model dynamically prioritizes highvalue, low-risk services, ensuring optimal bandwidth allocation while mitigating network failures. The risk function accounts for link failure probabilities and service priority levels, influencing path selection in real time. By enforcing strict bandwidth and survivability constraints, the ILP model strategically schedules fewer but more profitable flows, optimizing economic efficiency without compromising resilience. Simulation results demonstrate that our approach reduces service disruptions, minimizes high-risk routing, and improves revenue efficiency compared to Dijkstra-based routing. These findings offer a scalable framework for network operators to enhance disaster resilience while maintaining economic viability.

Federated learning (FL) has emerged as a powerful approach for privacy-preserving model training in autonomous vehicle networks, where real-world deployments rely on multiple roadside units (RSUs) serving heterogeneous clients with intermittent connectivity. While most research focuses on single-server or hierarchical cloud-based FL, multi-server FL can alleviate the communication bottlenecks of traditional setups. To this end, we propose an edge-based, multi-server FL (MS-FL) framework that combines performance-driven aggregation at each server—including statistical weighting of peer updates and outlier mitigation—with an application layer handover protocol that preserves model updates when vehicles move between RSU coverage areas. We evaluate MS-FL on both MNIST and GTSRB benchmarks under shard- and Dirichlet-based non-IID splits, comparing it against single-server FL and a two-layer edge-plus-cloud baseline. Over multiple communication rounds, MS-FL with the Statistical Performance-Aware Aggregation method and Dynamic Weighted Averaging Aggregation achieved up to a 20-percentage-point improvement in accuracy and consistent gains in precision, recall, and F1-score (95% confidence), while matching the low latency of edge-only schemes and avoiding the extra model transfer delays of cloud-based aggregation. These results demonstrate that coordinated cooperation among servers based on model quality and seamless handovers can accelerate convergence, mitigate data heterogeneity, and deliver robust, privacy-aware learning in connected vehicle environments.

This paper presents a novel multi-objective optimization (MOO) approach for managing risk in Software-Defined Networks (SDN) failure while concurrently considering operational cost and risk. Unlike traditional algorithms, we introduce an Ant Colony Optimization (ACO) algorithm adept at providing a range of nondominated solutions for path selection. Essential modifications have been made to promote efficient exploration, including penalizing the last link in a failed path and updating the pheromone matrix considering two objectives. Empirical testing with the ERNet network topology demonstrated the effectiveness of the proposed algorithm, identifying numerous nondominated and knee-point solutions. This work significantly contributes to network path optimization, providing diverse options for decision-makers, thereby enhancing the adaptability and dynamic management of traffic for large-scale failure scenarios in SDN.

Infotainment systems in vehicles have become important sources of digital evidence in forensic investigations. Analyzing data from these systems can provide valuable insights into a suspect's activities and interactions. In this paper, we propose a hybrid artificial intelligence (AI) framework that combines unsupervised learning using K-means clustering and language model analysis to enhance the forensic analysis process. The proposed methodology was applied to two distinct datasets from Hyundai and Mitsubishi infotainment systems. In the Hyundai dataset, the recall for contact names and phone numbers improved by 18% and 3% respectively when compared to clustering alone. Similarly, in the Mitsubishi dataset, the recall of song names improved by 2%. In addition, this hybrid approach enabled the discovery of more forms of forensically-relevant data stored in the infotainment systems, such as geographical locations and connected devices, that would have been infeasible to find with either manual analysis or clustering alone. Despite the presence of some hallucinations, the combination of these techniques resulted in improved ease of analysis and increased recall, demonstrating the potential of this hybrid approach in forensic investigations.

AI-based Network Intrusion Detection Systems (NIDS) provide effective mechanisms for cybersecurity analysts to gain insights and thwart several network attacks. Although current IDS can identify known/typical attacks with high accuracy, current research shows that such systems perform poorly when facing atypical and dynamically changing (polymorphic) attacks. In this paper, we focus on improving detection capability of the IDS for atypical and polymorphic network attacks. Our system generates adversarial polymorphic attacks against the IDS to examine its performance and incrementally retrains it to strengthen its detection of new attacks, specifically for minority attack samples in the input data. The employed attack quality analysis ensures that the adversarial atypical/polymorphic attacks generated through our system resemble original network attacks. We showcase the high performance of the IDS that we have proposed by training it using the CICIDS2017 and CICIoT2023 benchmark datasets and evaluating its performance against several atypical/polymorphic attack flows. The results indicate that the proposed technique, through adaptive training, learns the pattern of dynamically changing atypical/polymorphic attacks, identifies such attacks with approximately 90% balanced accuracy for most of the cases, and surpasses various state-of-the-art detection and class balancing techniques.

Agile network security is paramount in our modern world which is currently dominated by Internet systems and expanding digital spaces. This rapid digital transformation has created more opportunities for cyberattackers to exploit different vulnerabilities and launch sophisticated and continuously evolving cyberattacks. Increasingly, intrusion detection systems are relying on new methods based on Machine Learning (ML) and Deep Learning (DL) techniques to detect and mitigate such cyberattacks. While such techniques normally can identify known network attack patterns with a reasonable degree of success, their ability to identify complicated atypical, polymorphic, and unknown attacks is shown to be limited. In this paper, we present a comprehensive survey of recent research for detecting unknown, atypical, and polymorphic network attacks using DL techniques. We further highlight and discuss the main challenges in this area and identify the future research directions.

Federated Learning (FL) protects privacy during autonomous vehicle machine learning (ML) operations. FL enables cooperative training of a single ML model across multiple edge devices, leveraging distributed datasets while maintaining data locality. Although much research has concentrated on single-server FL for autonomous driving applications within vehicular networks, real-world scenarios often involve several concurrent servers capable of benefiting from each other's knowledge. However, these servers' trustworthiness is paramount when using their global models, as an imprudent choice could significantly decrease FL performance and accuracy. In this paper, we introduce a novel trust-based knowledge-sharing approach among FL servers, wherein the accuracy of shared global models on clients' local data serves as the trust metric. Our proposed methodology enables servers to utilize shared global models from reliable servers for their clients, thereby improving training accuracy and reducing loss. This enhancement is particularly notable during the initial training rounds compared to base FL implementation.

Network lifetime and localization are critical design factors for a number of wireless sensor network (WSN) applications. These networks may be randomly deployed and left unattended for prolonged periods of time. This means that node localization is performed after network deployment, and there is a need to develop mechanisms to extend the network lifetime since sensor nodes are usually constrained battery-powered devices, and replacing them can be costly or sometimes impossible, e.g., in hostile environments. To this end, this work proposes the energy-aware connected k-neighborhood (ECKN): a joint position estimation, packet routing, and sleep scheduling mechanism. To the best of our knowledge, there is a lack of such integrated solutions to WSNs. The proposed localization algorithm performs trilateration using the positions of a mobile sink and already-localized neighbor nodes in order to estimate the positions of sensor nodes. A routing protocol is also introduced, and it is based on the well-known greedy geographic forwarding (GGF). Similarly to GGF, the proposed protocol takes into consideration the positions of neighbors to decide the best forwarding node. However, it also considers node residual energy in order to guarantee the forwarding node will deliver the packet. A sleep scheduler is also introduced in order to extend the network lifetime. It is based on the connected k-neighborhood (CKN), which aids in the decision of which nodes switch to sleep mode while keeping the network connected. An extensive set of performance evaluation experiments was conducted and results show that ECKN not only extends the network lifetime and localizes nodes, but it does so while sustaining the acceptable packet delivery ratio and reducing network overhead.

Today's networked systems face significant security challenges due to sophisticated attacks. Several Machine Learning (ML) and Deep Learning (DL) models are employed to combat these diverse attacks. Adversarial attacks, which can evade detection by AI-based intrusion detection systems (IDS) through small alterations to network attack traffic, pose a significant concern. These AI-synthesized adversarial attacks must adhere to network constraints to seem plausible. In this work, we explore the validation criteria for such adversarial attacks and propose a methodology for analyzing their quality. We evaluate adversarial attack samples synthesized by state-of-the-art generative DL models such as Variational autoencoder (VAE), Conditional Variational autoencoder (CVAE), Generative Adversarial Network (GAN) and compare the performance with our CVAE-Adversarial Network (CVAE-AN) model. Results indicate the effectiveness of CVAE-AN in synthesizing realistic adversarial attacks.

The growth of electric vehicle (EV) adoption is bringing an increased demand for electric vehicle supply equipment (EVSE) infrastructure. With this growth, however, it is inevitable that vulnerabilities are discovered, which motivates an in-depth analysis of the security posture of EVSE infrastructure and development of a strong cybersecurity program to ensure its security and resiliency. In this research paper, we will analyze and highlight various vulnerabilities in EVSE systems, including weak authentication mechanisms, and end-to-end communications. We will identify as well potential attack scenarios that can create vulnerabilities that have consequences that range from physical damage to major service disruptions, and identify various solutions and countermeasures to mitigate these vulnerabilities.

The increasing popularity and usage of unmanned aerial vehicles (UAVs) has brought about new challenges in airspace management. With the number of drones expected to grow even further in the coming years, there is an urgent need for an autonomous traffic management system (TMS) that can safely and effectively manage drone traffic in the airspace. It is critical that this TMS be built with principles of the Confidentiality, Integrity, and Availability (CIA) triad. In this paper, a traffic management system for UAVs is presented that takes advantage of a Hyperledger Fabric blockchain network. The TMS provides a decentralized and secure method to manage and deconflict drone flight paths, allowing for safe navigation in crowded airspaces. Through a series of simulated experiments, we demonstrated the system's capabilities in handling path creation, multiple conflict resolutions, and large numbers of drones. Simulated tests showed that the proposed system was able to handle deconfliction of 1000 drones inside of a one square kilometer, and returned calculated paths for drones in 60 to 2000 ms with up to 100 deconflictions. The Hyperledger Fabric powered traffic management system showcased the potential to leverage permissioned blockchain technology in improving drone traffic management.

Unmanned Aerial Vehicles (UAVs) are seeing increased use in critical operations for law enforcement, military, industrial control surveillance and more. These hostile operating environments combined with the UAVs reliance on wireless protocols produces an increased threat level. Many attacks against the UAV are becoming commonplace as they are simple to conduct with inexpensive hardware, such as spoofing and jamming. Unfortunately, as many of these vulnerabilities exist within underlying technologies, securing the UAV becomes a difficult task. A promising approach to identifying and mitigating these attacks is the development of an intelligent intrusion detection system (IDS). The proposed approach uses principal component analysis (PCA) and one-class classifiers to detect attacks. This allows for the use of flight logs for training data, providing a versatile and ubiquitous approach. The proposed detection method is integrated into a fully developed IDS called MAVIDS. This IDS operates onboard the UAV within a resource-constrained agent device, allowing it to detect and potentially mitigate attacks even when communication to the ground control station is lost from jamming. The approach shows to be effective against GPS spoofing and jamming with macro averaged F1 scores of 90.57% and 94.3% respectively.

Artificial intelligence is a rapidly developing field of research with many practical applications. Congruent to advances in technologies that enable big data, deep learning, and neural networks to train, learn, and predict, artificial intelligence creates new risks that are difficult to predict and manage. Such risks include economic turmoil, existential crises, and the dissolution of individual privacy. If unchecked, the capabilities of artificially intelligent systems could pose a fundamental threat to privacy in their operation or these systems may leak information under adversarial conditions. In this article, we survey the literature and provide various scenarios for the use of artificial intelligence, highlighting potential risks to privacy and offering various mitigating strategies. For the purpose of this research, a North American perspective of privacy is adopted. Impact statement—While an appreciation of the privacy risks associated with artificial intelligence is important, a thorough understanding of the assortment of different technologies that comprise artificial intelligence better prepares those implementing such systems in assessing privacy impacts. This can be achieved through the independent consideration of each constituent of an artificially intelligent system and its interactions. Under individual consideration, privacy-enhancing tools can be applied in a targeted manner to reduce the risk associated with specific components of an artificially intelligent system. A generalized North American approach to assess privacy risks in such systems is proposed that will retain applicability as the field of research evolves and can be adapted to account for various sociopolitical influences. With such an approach, privacy risks in artificial intelligent systems can be well understood, measured, and reduced.

Artificial Intelligence (AI) techniques provide effective solutions for the detection of many aberrant network traffic patterns and attack flows. However, the validation of these techniques often relies on one training dataset. Recent results show that such training may fail in the face of dynamically-changing cyberattacks. Given the increased sophistication of cyberattacks nowadays, it is imperative to examine and improve the performance of such AI models. This paper proposes a defensive AI engine combined with a twofold feature selection technique and hyperparameter optimization of the AI model. In this work, we utilize the proposed system for binary attack flow identification and the AI models are trained and validated on the CICIDS2017 dataset. The system is then evaluated using synthesized atypical attack flows to mimic real-world scenarios. We demonstrate the effectiveness of the proposed atypical attack flow detection approach using several Deep Learning and Machine Learning models including DNN, Linear-SVC, and Stacked Decision Tree Classifier (S-DTC). Simulation results demonstrate that the proposed defensive AI engine significantly improves the True Positive Rate (TPR) of AI models on multiple atypical attacks.

Network Intrusion Detection Systems (NIDS) are powerful tools for identifying and deterring cybersecurity attacks nowadays. However, while these modern IDS can detect typical attacks, recent studies show their poor performances in identifying unknown or dynamically changing atypical attacks. Another issue with the training aspect of such systems is the problem of class imbalance which impedes their performance, especially for minority attack classes. This renders IDS systems vulnerable to both adversarial as well as non-AI synthesized atypical attacks when deployed in a real network. To reduce misclassification (especially for minority classes) and detect atypical attack flows, we propose a novel adversarial incremental learning approach based on a hybrid model consisting of a Conditional Variational Autoencoder (CVAE) and a Generative Adversarial Network (GAN) namely, CVAE-Adversarial Network (CVAE-AN). The binary IDS has been trained using the CICIDS2017 dataset and evaluated using multiple atypical attacks. Simulation results demonstrate that the proposed technique significantly improves the performance of the IDS against different atypical attacks and outperforms the state-of-the-art detection models as well as class balancing methods.

Distributed computing systems are increasing in popularity and being widely used as a new way of large-scale data processing. However, to achieve a reliable and efficient performance in a distributed environment, it is important to deal with system anomalies as soon as they are encountered. In this paper, two novel anomaly detection algorithms will be introduced and compared with previous anomaly detection algorithms. These novel algorithms are devised based on data summarization and error prediction in comparison with previously extracted data. The result of our experiments show that the proposed methods exhibit higher performance in terms of precision and accuracy.

Driving simulators play an important role in vehicle research. However, existing virtual reality simulators do not give users a true sense of presence. UniNet is our driving simulator, designed to allow users to interact with and visualize simulated traffic in mixed reality. It is powered by SUMO and Unity. UniNet’s modular architecture allows us to investigate interdisciplinary research topics such as vehicular ad-hoc networks, human-computer interaction, and traffic management. We accomplish this by giving users the ability to observe and interact with simulated traffic in a high fidelity driving simulator. We present a user study that subjectively measures user’s sense of presence in UniNet. Our findings suggest that our novel mixed reality system does increase this sensation. Author

As over 95% of IT security breaches result from human error, there is a need to design a system to detect and prevent it. In this paper, we propose a human error prevention system that examines the users’ eye gaze behavior and determines if a human error is likely to occur. 13 expert network administrators and 18 non-experts were selected for this study. The results confirm the ability to detect human error using eye gaze data with 99.74%, 99.47%, 99.23% accuracy using KStar, Random Forest and J48 classifiers.

In this thesis, we propose a functionality-aware system to evaluate and recommend mobile applications with security and privacy awareness. The proposed system has a security layer that evaluates an application and classifies it as being malign or benign. In this way, only applications classified as benign are considered for the functionality-aware recommendation. Also, we employ a technique, called Logical Predicate Mapping (LPM), which allows users to understand the permissions and API calls requested by the app, as well as privacy risks. This information is grouped with other metrics retrieved such as popularity, usability and privacy and shown to users. This way they can decide what to do and understand what can happen.