The $6^{th}$ generation (6G) of mobile networks blazes the trail towards an intelligent information society where smart applications and services with intensified requirements will be offered to users. In this complex environment, key advancements are necessary in the development of innovative resource management strategies that can lead to a robust network operation. In this work, we exploit the unified computational configuration of a converged optical-wireless cell free-based 6G network in order to introduce a load offloading mechanism targeting on the efficient management of the system's computational resources. We also propose a mathematical framework to evaluate the performance of the offloading mechanism and we assess its effectiveness under different resource management policies. It is shown that the incorporation of the offloading procedure generally improves the system's service provisioning in a cost-effective manner.

Autonomous Vehicles (AVs) rely on real-time processing of natural images and videos for scene understanding and safety assurance through proactive object detection. Traditional methods have primarily focused on 2D object detection, limiting their spatial understanding. This study introduces a novel approach by leveraging 3D object detection in conjunction with augmented reality (AR) ecosystems for enhanced real-time scene analysis. Our approach pioneers the integration of a synthetic dataset, designed to simulate various environmental, lighting, and spatiotemporal conditions, to train and evaluate an AI model capable of deducing 3D bounding boxes. This dataset, with its diverse weather conditions and varying camera settings, allows us to explore detection performance in highly challenging scenarios. The proposed method also significantly improves processing times while maintaining accuracy, offering competitive results in conditions previously considered difficult for object recognition. The combination of 3D detection within the AR framework and the use of synthetic data to tackle environmental complexity marks a notable contribution to the field of AV scene analysis.

It is essential to create effective strategies for safeguarding private data while preserving the value of textual data in the face of rising privacy concerns in the digital era. In this research, we give a thorough investigation of Conditional Random Fields (CRF), Long Short-Term Memory (LSTM), and Embeddings from Language Models (ELMo)-based text anonymisation methods [1]. The strengths of CRF, LSTM, and ELMo are combined in the proposed approach to overcome the difficulty of text anonymisation. LSTM is a recurrent neural network architecture that can model long-term dependencies, CRF is a probabilistic graphical model that captures the dependencies among the sequence of words, and ELMo offers contextual word representations utilizing deep bidirectional language models. These models are leveraged in order to identify and subsequently obfuscate sensitive information in the frame of textual anonymisation. Experimental results show that when compared to conventional methods, all three methods—CRF, LSTM, and ELMo—achieve considerable improvements in text anonymisation.

Swarms of cooperative Unmanned Aerial Vehicles (UAVs) have emerged as a new promising solution for accomplishing complex and large scale tasks, that would be challenging or even impossible for a single UAV. Coordinating multiple UAVs towards a common mission is quite challenging, as multiple operational factors must be ensured in terms of flight reliability, safety, resource allocation and ubiquitous communications. This paper presents an integrated UAV swarm management platform for autonomous area and infrastructure inspection and proposes a novel network architecture for relaying swarm communications in underserved areas. Real trials and measurements were performed on the field, providing preliminary results and proof-of-concept insights.

We present novel results for the uncoded and coded bit-error probability (BEP) of optically pre-amplified pulse-position modulation (PPM) wireless systems. For uncoded systems, a novel analytic method for the evaluation of the BEP is derived. The method takes into account the non-ideal optical filter response and utilizes a finite Karhunen-Lo & egrave;ve series expansion to calculate the BEP. Using the proposed approach, it is possible to accurately evaluate the PPM BEP for arbitrarily shaped filters where the well-established chi(2) method only provides approximate results. Considering a Lorentzian filter response, the discrepancy between the two methods amounts to 0.5 dB in a variety of filter bandwidths and PPM modulation orders. The Lorentzian filter response was chosen as an illustrative practical example whose series can be calculated analytically. The proposed method is also valid for any type of optical filter for which the Karhunen-Lo & egrave;ve series expansion can be calculated analytically or numerically. Due to the finite number of terms that are required irrespective of the signal energy level, the proposed method can also be applied without loss of accuracy to assess the system performance under the effects of turbulence and adverse weather conditions. For coded systems with Lorentzian filters, Monte-Carlo simulations are utilized to evaluate the BEP performance of the 5G LDPC codes, and it is demonstrated that they impart an energy gain up to 3.3 dB for 4-PPM and 2.3 dB for 16-PPM at a target BEP of 10 -5 . The optimal code rates are also discussed for several combinations of the optical filter bandwidth and PPM modulation order and it is shown that in almost all of the cases the optimal code rate is 11/13. Moreover, the sum-product and min-sum decoders perform within 0.1 dB from each other for the best code rates, which points towards the utilization of the min-sum decoder in all settings, since its operation does not require knowledge of the filter parameters. Finally, the comparison between the coded systems with Lorentzian and ideal passband filters exhibits the same 0.5 dB discrepancy that was observed for uncoded systems.

In this paper, we evaluate the performance of a vehicular ad hoc network that enables Vehicle-to-Everything (V2X) communication considering direct and indirect packet transmission. As a performance indicator, we use the blocking probability due to the unavailability of resources. For the blocking probability calculation, we utilize two loss models from the teletraffic theory. The results demonstrate that both the capacity of the V2X links and the number of hops in the multi-hop transmission significantly affect the blocking probability of a request for service and consequently the QoS.

With the ongoing integration of machine learning models into critical infrastructure, the resilience of these systems against adversarial attacks is important for all domains. This paper introduces an adversarial attack generator framework against a network dataset that is part of OCPP Dataset using CI-CFlowMeter parser. We conduct a comprehensive evaluation of various prominent adversarial attacks, including FGSMA, JSMA, PGD, C&W, and more to assess their efficacy on the OCCP dataset. The Adversarial Generator is meticulously evaluated, demonstrating a significant impact in the models performance to detect potential perturbations. The results showcased the impact of the different type of adversarial attacks, contributing to a critical advancement in future defense strategies that need to be utilised in order to protect industrial control systems.

The evolution towards the emergence of the smart 6 th generation of telecommunication networks targets on un-precedentedly transforming the existing network systems. This unparallel transformation relies on being able to serve in zero-latency highly demanding applications and services hosted on a massive number of devices. This new era of telecommunication networks has designated the need to develop innovative solutions that can intelligently manage the systems' resources in order to deal with the intensified requirements. In this work, we propose a load offloading mechanism aiming to efficiently manage the computational resources of a cell free-based 6G network. We also present a novel traffic-engineering model that aims to evaluate the proposed offloading mechanism and we validate the analytical model with numerical simulation results in order to demonstrate its high accuracy. Finally, we examine the system's performance for different parameter values and it is shown that the offloading mechanism has a strong impact on improving the system's service provisioning.

Autonomous driving systems consist of vehicles that are able to communicate not only with other vehicles but also with entities in their environment, forming vehicle-to-everything (V2X) communication. However, the V2X applications have intense requirements posing a significant challenge to the telecommunication infrastructure. In this work, we consider two types of transmission, i.e. direct and indirect, and we utilize analytical traffic-engineering models with a view to conduct a performance analysis of a vehicular network that enables V2X communication. We additionally propose two resource management strategies in order to decrease the request rejection probability and consequently ensure enhanced communication conditions. The results reveal that the proposed resource management strategies constitute a strong asset for improving the system’s service provisioning capability.

Telecommunications profoundly impacts all major aspects of our everyday life. As a consequence, student instruction typically includes a series of specialized courses, each addressing a distinct telecommunication area, separating wireless from fixed (optical) communications. This creates the problem of knowledge fragmentation, hindering the student’s perception of the topic since, at the service level, the applications and services offered to the users seem “virtually” independent from the underlying infrastructure. In this paper, to address this problem, we designed, analyzed, and implemented a 6 h course module on the five generations of wireless and fixed networks, which was presented as an integral part of the undergraduate course “Broadband Communications”, which was offered at the Dept. of Electrical and Electronic Engineering, School of Pedagogical and Technological Education (ASPETE), Athens, Greece. The main targets of this module are the following. Firstly, it aims to familiarize students with the fixed generations taxonomy, defined by the ETSI Industry Specification Group (ISG) F5G. This taxonomy serves as a foundation for understanding the evolution of telecommunications technologies. Secondly, the module seeks to integrate the acquired knowledge of the students in their previous telecommunication-related courses. During their curriculum, this knowledge was divided into two separate parts: wireless and fixed (optical). By coupling these two areas, students can develop a deeper understanding of the field. Lastly, the module aims to explore cutting-edge technologies and advancements in the telecommunications industry. In this way, it prepares students to enter the professional world during the fifth-generation era. Additionally, it provides them with valuable insights into the ongoing research and development in the field of 6G. Overall, this module serves as a comprehensive platform for students to enhance their understanding of telecommunications, from the foundational concepts to the latest advancements. To evaluate the impact of this module, the students were asked to fill out a questionnaire that included seven questions upon module completion. This questionnaire was completed successfully by 32 students in the previous academic year and by 16 students in this academic year. Moreover, a 20-question multiple choice quiz was offered to the students, allowing us to probe more into the typical errors and misconceptions about the topic.

The increased availability of medical data has significantly impacted healthcare by enabling the application of machine / deep learning approaches in various instances. However, medical datasets are usually small and scattered across multiple providers, suffer from high class-imbalance, and are subject to stringent data privacy constraints. In this paper, the application of a data regularization algorithm, suitable for learning under high class-imbalance, in a federated learning setting is proposed. Specifically, the goal of the proposed method is to enhance model performance for cardiovascular disease prediction by tackling the class-imbalance that typically characterizes datasets used for this purpose, as well as by leveraging patient data available in different nodes of a federated ecosystem without compromising their privacy and enabling more resource sensitive allocation. The method is evaluated across four datasets for cardiovascular disease prediction, which are scattered across different clients, achieving improved performance. Meanwhile, its robustness under various hyperparameter settings, as well as its ability to adapt to different resource allocation scenarios, is verified.

In the realm of data privacy, the ability to effectively anonymise text is paramount. With the proliferation of deep learning and, in particular, transformer architectures, there is a burgeoning interest in leveraging these advanced models for text anonymisation tasks. This paper presents a comprehensive benchmarking study comparing the performance of transformer-based models and Large Language Models(LLM) against traditional architectures for text anonymisation. Utilising the CoNLL-2003 dataset, known for its robustness and diversity, we evaluate several models. Our results showcase the strengths and weaknesses of each approach, offering a clear perspective on the efficacy of modern versus traditional methods. Notably, while modern models exhibit advanced capabilities in capturing contextual nuances, certain traditional architectures still keep high performance. This work aims to guide researchers in selecting the most suitable model for their anonymisation needs, while also shedding light on potential paths for future advancements in the field.

We present a novel analytical method for the evaluation of the Bit-Error-Probability (BEP) of optically pre-amplified Pulse-Position-Modulation (PPM) receivers. The method takes into account the non-ideal optical filter response and utilizes a finite series expansion to calculate the BEP. Using the proposed approach, it is possible to accurately evaluate the PPM BEP for arbitrarily shaped filters where the well-established χ 2 method only provides approximate results. The method is applied for ideal passband and Lorentzian response wide optical filters, and the obtained results are in perfect agreement with the χ 2 approximation for the ideal filter, while a discrepancy of 0.5 dB is observed between the two methods for the Lorentzian filter.

Autonomous Vehicles (AVs) stand as the vanguard of the automotive industry's evolution, offering a multitude of advantages in terms of transportation efficiency and applications of critical importance. Notably, their interconnection with various smart devices, such as smartphones and associated services, is achieved effortlessly. However, these merits are counterbalanced by significant security risks pertaining to human safety and the potential exposure of personal data. This work introduces SiHoneypot, an innovative honeypot system rigorously crafted to address security challenges intrinsic to AVs. SiHoneypot leverages Digital Twins and incorporates state-of-the-art trends in software deployment, providing a faithful emulation of Autonomous Vehicle systems. Demonstrating its efficacy as a strategic decoy, SiHoneypot affords sufficient time for other security systems to enact responsive measures. Experimental results underscore the minimal resources required for the deployment of SiHoneypot, emphasizing its operational efficiency and resource optimization. Moreover, the inherent extensibility and versatility of SiHoney-pot's architecture are showcased, illustrating its adaptability to evolving security challenges within the dynamic landscape of autonomous vehicular technologies.

In the ever-evolving era of Artificial Intelligence (AI), model performance has constituted a key metric driving innovation, leading to an exponential growth in model size and complexity. However, sustainability and energy efficiency have been critical requirements during deployment in contemporary industrial settings, necessitating the use of data-efficient approaches such as few-shot learning. In this paper, to alleviate the burden of lengthy model training and minimize energy consumption, a finetuning approach to adapt standard object detection models to downstream tasks is examined. Subsequently, a thorough case study and evaluation of the energy demands of the developed models, applied in object detection benchmark datasets from volatile industrial environments is presented. Specifically, different finetuning strategies as well as utilization of ancillary evaluation data during training are examined, and the trade-off between performance and efficiency is highlighted in this low-data regime. Finally, this paper introduces a novel way to quantify this trade-off through a customized Efficiency Factor metric.

In this paper, we consider a vehicle that has an access point of fixed capacity. The vehicle accommodates a finite number of users who generate calls (quasi-random process). Each call requires a single bandwidth unit to be serviced for an exponentially distributed service time. The applied call admission policy depends on which phase the vehicle is. Two phases are taken into account: a moving and a stop one. During the moving phase, only new calls are serviced under the complete sharing policy. During the stop phase, both new and handover calls are being serviced, while the latter are being prioritized utilizing a probabilistic bandwidth reservation policy. The analytical determination of the various performance metrics of the system presented in this paper is accurate and is based on three-dimensional Markov chains.

The revolution of Artificial Intelligence (AI) has brought about a significant evolution in the landscape of cyberattacks. In particular, with the increasing power and capabilities of AI, cyberattackers can automate tasks, analyze vast amounts of data, and identify vulnerabilities with greater precision. On the other hand, despite the multiple benefits of the Internet of Things (IoT), it raises severe security issues. Therefore, it is evident that the presence of efficient intrusion detection mechanisms is critical. Although Machine Learning (ML) and Deep Learning (DL)-based IDS have already demonstrated their detection efficiency, they still suffer from false alarms and explainability issues that do not allow security administrators to trust them completely compared to conventional signature/specification-based IDS. In light of the aforementioned remarks, in this paper, we introduce an AI-powered IDS with explainability functions for the IoT. The proposed IDS relies on ML and DL methods, while the SHapley Additive exPlanations (SHAP) method is used to explain decision-making. The evaluation results demonstrate the efficiency of the proposed IDS in terms of detection performance and explainable AI (XAI).

The evolution of the telecommunication networks towards their 6th generation has emerged new research directions in order to deal with the challenges that are posed by the high-complexity of the new infrastructure. In this new era, the structuring of the network management infrastructure is a complex endeavor that should efficiently control both communication and computational resources. In this paper, we present a set of strategies for managing the computational load in a cell-free-based, converged optical-wireless 6G network. The proposed approaches target to control the computational load of a cell-free network, by either compressing the load by considering load-thresholds, or offloading the load to the SDN controller of the fixed network. Both strategies are mathematically formulated by considering traffic-engineering formulas, while their performance is evaluated by comparing analytical results with corresponding results from a baseline scenario, where no load control strategies are applied. Moreover, the proposed analysis can be applied in order to determine the capacity of the network controllers that is required in order to guarantee pre-determined Quality of Service requirements.

The authors study and evaluate a mobility-aware call admission control algorithm in a mobile hotspot. More specifically, a vehicle which has an access point of a fixed capacity and may alternate between stop and moving phases is considered. In the stop phase, the vehicle services new and handover calls. To prioritise handover calls a probabilistic bandwidth reservation policy is considered where a fraction of the capacity is reserved for handover calls. Based on this policy, new calls may enter the reservation space with a predefined probability. In addition, handover calls have the option to wait in a queue of finite size if there are no available resources at the time of their arrival. In the moving phase, the vehicle services only new calls under the classical complete sharing policy. In both phases, calls arrive in the system according to a quasi-random process, require a single bandwidth unit for their acceptance in the system and have an exponentially distributed service time. To analytically determine the various performance measures, such as time congestion probabilities, call blocking probabilities and link utilisation, an accurate analytical method is presented based on three-dimensional Markov chains.

The digitisation of the smart electrical grid provides several advantages and valuable services, such as self-monitoring, pervasive control and smart healing. However, despite the benefits of this progression, critical cybersecurity and privacy issues are raised due to the vulnerabilities of legacy Electrical Power and Energy Systems (EPES) and the evolution of stealthy cyberthreats and malware. In this paper, we give emphasis to False Data Injection Attacks (FDIAs) that can affect the EPES State Estimation (SE). In particular, we investigate two FDIA categories, namely: (a) Global Positioning System (GPS) Spoofing Attacks and (b) IEEE C37.118 FDIAs against an actual testbed emulating a high-voltage IEEE 9-Bus transmission grid. Finally, we provide a relevant Intrusion Detection System (IDS) capable of detecting the aforementioned FDIAs. The evaluation analysis demonstrates the impact of the above FDIAs and the efficiency of the proposed IDS.