Despite being a cornerstone of scientific progress, reproducibility remains a key challenge in deep reinforcement learning (DRL). This paper presents a reproducibility study of the RELIANCE method, a DRL-based approach for drone collision avoidance originally proposed by Ouahouah et al. We identify critical flaws in RELIANCE’s exploration strategy and reward function, which impede learning efficiency and task completion. By refining the exploration strategy and introducing a distance-based reward shaping mechanism, we achieve notable improvements in collision avoidance performance, reward accumulation, and success rates. Our modifications yield a 17-fold increase in success rates and a 15.56% reduction in collisions, alongside higher cumulative rewards. These findings highlight the vital role of early exploration and carefully crafted reward functions in DRL applications.
To address rising Quality of Service (QoS) requirements in beyond 5G systems, this paper proposes a data-centric approach for user plane QoS control in beyond 5G systems. Building upon the PSBA-XG platform, previously designed for control plane optimization, we extend its capabilities to monitor and manage the User Plane Function (UPF) performance. By aggregating standard-compliant usage reports using a publish/subscribe architecture, our solution enables rapid detection of QoS degradation and reconfiguration of user sessions. Experimental results on an open-source 5G testbed demonstrate that our approach allows for more flexible QoS management, notably rerouting user traffic to edge resources in case of congestion, than a plain 5G system, without introducing additional complexity. These results highlight the potential of data-centric architectures to enhance QoS in future mobile network systems.
Ransomware has evolved into a full-blown, multi-stage, persistent threat attacking the very core of our critical infrastructures, from cloud to IoT. Although many strategies have been proposed for defense, there is a lack of a comprehensive view on how they relate to the development process and integrate recent advancements. In this paper, we provide an organized overview of ransomware defense by thoroughly examining peer-reviewed articles published between 2018 and 2025 from reputable venues (e.g., IEEE, ACM). We propose a comprehensive taxonomy for detection techniques, encompassing static, behavioral, ML-based, and pre-encryption methods. Our analysis compares real-life deployment challenges such as dataset scarcity and model generalizability against emerging trends such as fileless and browser-based attacks. By identifying significant research gaps in adversarial robustness and explain ability, this survey outlines an actionable agenda to develop scalable, privacy-preserving defenses against ransomware as next-generation defenses continue to evolve.
Vehicular visible light communication (V VLC) systems, when combined with reconfigurable intelligent surfaces (RIS), present promising opportunities for improving communication reliability and efficiency in vehicle to vehicle (V2V) environments. Nevertheless, safeguarding these systems at the physical layer remains a critical challenge, particularly given their exposure to jamming threats. In this study, we investigate the physical layer security performance of RIS assisted V2V VLC systems under jamming scenarios, employing realistic V2V VLC channel models. We develop a methodology to examine the impact of various security strategies in mitigating the adverse effects of jamming. Our analysis examines key parameters including the signal to noise ratio SNR, the secure communication rate and the count of RIS units. Simulation results confirm that the proposed security systems significantly enhance the resilience of V2V VLC networks in the presence of jamming attacks. These results offer useful perspectives for the reliable design structure and deployment of RIS based V2V VLC systems in practical vehicular communication settings.
Despite years of research and the introduction of advanced features in 5G systems, no commercially available 5G device has yet met the stringent latency and reliability requirements defined for ultra-reliable and low-latency communication (URLLC). This raises fundamental questions about whether and how URLLC can be practically implemented under current technological constraints. To answer this question, in this work, we explore the combination of several features defined by the 3GPP standardization, including link adaptation, retransmission schemes, and scheduling algorithms. We evaluate the performance of these techniques to investigate the extent to which the requirements of URLLC can be achieved by using the NS-3 5G LENA simulator. Our results demonstrate that the shortest latency at which a reliability level of 99.999% can be achieved is 4 milliseconds, provided that the wireless channel is of exceptionally high quality. This level of performance is enabled by combining a retransmission scheme with an effective scheduling algorithm. However, in the presence of poor channel quality, it becomes necessary to relax the reliability constraint and/or provide more resources to the transmission to maintain system feasibility.
To ensure unmanned aerial vehicle (UAV) swarm safety in areas lacking satellite navigation, accurate localization must be achieved through alternative methods such as relative positioning and clustering. A key challenge involves balancing higher localization accuracy with lower computational complexity, especially in urban environments. Additionally, maintaining up-to-date data transmission to ground stations is critical for real-time localization in dynamic scenarios.This paper proposes a cooperative relative localization scheme based on clustering for UAV swarms. The proposal employs the degree centrality metric to form clusters and connectivity information for cluster head selections, and balances intra-cluster cooperation with inter-cluster packet loss, ensuring the freshness of transmitted data.A first evaluation of the proposed scheme allows us to venture that it will achieve better performance (in terms of mean square error) compared to state-of-the-art techniques for relative localization of UAV swarms.
Named Data Networking (NDN), as an emerging architecture for evolving IP networks, focuses on content names rather than their origins. It distinguishes itself through key features such as one-Interest/one-Data transfer, in-network caching, Interest aggregation, and Data security. However, despite these advantages, NDN lacks a QoS mechanism to differentiate between data types, especially with the increasing heterogeneity and scale of network traffic. The rapid growth of applications such as IoT, VANETs, and e-health systems highlights the urgent need for effective QoS strategies. Existing state-of-the-art QoS strategies often fail to capture the dynamic nature of network traffic, while traditional IP-based QoS mechanisms cannot be directly applied to NDN due to its architectural specificity. To address this challenge and enable service differentiation, we propose MPQ-NDN: a QoS strategy based on dynamic packet prioritization according to their generation frequency, predicted using a Markov model. This prediction enables proactive adaptation to traffic variations and extends NDN forwarding with priority queues. Simulation results conducted in ndnSIM demonstrate that MPQ-NDN significantly improves network performance, responsiveness, and user satisfaction.
This paper proposes a 5G network data space targeting the collection, storing, and sharing of metrics generated by 5G infrastructure, such as latency, throughput, and network slice performance, for universal access by applications. The primary challenge is standardizing these metrics across commercial vendors (e.g., Nokia, Ericsson, Huawei, Samsung) and open-source implementations to ensure interoperability in a decentralized ecosystem. Current approaches, such as NWDAF, O-RAN, and Gaia-X, face this interoperability challenge but remain limited in scope. We present a vendor-agnostic framework that supports multiple metric-gathering methods, including standardized mechanisms, and proprietary telemetry protocols, to enable cooperative data collection The paper includes a detailed architecture which is organized into four layers—Data Collection, Standardization, Data Management, and Application Interface, providing solutions to vendor heterogeneity, fostering scalable, secure, and interoperable data sharing for advanced applications and services. This work is a conceptual and architectural foundation, that outlines the main directions for such a future 5G Dataspace, with validation and prototyping left to subsequent research.
The use of Unmanned Aerial Vehicles (UAVs), or drones, has become increasingly widespread across a variety of applications, ranging from filmmaking and surveillance to rescue operations, espionage, and military missions. This proliferation has given rise to drone networks, swarms, fleets, and the Internet of Drones (IoD). As these drones are used to carry confidential and sensitive missions, location privacy and security are of essential importance to ensure. Drones are required to broadcast location data and identifiers for navigation and safety, yet they inherit the resource constraints of the Internet of Things (IoT), such as limited battery life and computational capacity, which necessitate lightweight and efficient security solutions. To address these challenges, we propose an identifier-changing scheme designed to prevent tracking and safeguard both identity and location privacy. Our approach relies on inter-drone cooperation, coordinated silence periods, and location obfuscation to achieve unlinkability and anonymity. Techniques inspired by vehicular networks, when adapted to the energy limitations of drones, show strong potential in enhancing privacy within the IoD.
Wireless Sensor Networks (WSNs) are critical for applications requiring accurate environmental monitoring and dependable performance. A major challenge in these networks is the occurrence of coverage holes—unmonitored regions that degrade reliability and efficiency. This paper presents a geometry-driven hole detection framework in which sensors are modeled as uniform disks, overlapping areas are merged to form a unified coverage region, and the convex hull is computed to define the network boundary. Subtracting the coverage region from the hull exposes genuine coverage gaps. To restore coverage, we propose an energy-aware Quality-of-Service (QoS) scheduler that prioritizes hole healing according to sensor battery levels and hole criticality, thereby extending network lifetime. Two complementary algorithms are introduced: a lattice-based sensor placement strategy that ensures minimum 1-coverage and a priority scheduling mechanism that determines the order of hole restoration. Simulation results demonstrate that the proposed framework achieves high precision in hole detection, reduces the number of sensors required for recovery, and significantly improves energy efficiency, ultimately enhancing both reliability and performance in WSNs.
Efficient random access can be used in scenarios with a massive number of IoT devices. Among modern random access protocols, Irregular Repetition Slotted ALOHA (IRSA) offers excellent asymptotic performance (for large frame sizes), but its finite-frame efficiency is lower and difficult to optimize analytically. In this work, we introduce limited mid-frame feedback to better coordinate users and improve performance: FeedbackIntegrated Two-phase IRSA (FIT-IRSA). We formulate IRSA with feedback as a deep reinforcement learning (DRL) problem. Using policy gradient methods, we learn transmission strategies that improve throughput under varying loads, as demonstrated in our simulation results. This provides a practical alternative to classical density-evolution-based optimization, which applies mainly to large frames.
The Internet of Vehicles (IoV) integrates smart vehicles equipped with sophisticated communication systems for secured and improved seamless urban mobility. Nonetheless, there are obstacles to the effective provision of services in these environments due to the heterogeneity of resource capabilities, varying application demands, and the rapid change in vehicular networks. To meet these issues, we have developed a distributed four-layer Vehicular Fog Computing (VFC) architecture with an emphasis on task management and latency reduction. Digital Twins (DTs) facilitate the instant virtualization of vehicles, fog/edge, and cloud resources, which aids in monitoring, predicting, and controlling cross-layer coordination. Deep Reinforcement Learning (DRL) is used to facilitate autonomous decision-making on resource allocation and task offloading. The DRL agent operates based on simulated DTs and learns workload distribution to address the dynamic network and QoS thresholds. Our DT-DRL framework outperforms conventional solutions in the aspects of service quality, adaptability, and scalability in dynamic IoV environments.
The realization of sustainable logistics and regional transportation services requires the deployment of automated driving technologies, which in turn demands further latency reductions in mobile networks. However, one of the major challenges in existing mobile networks, including 5G, is addressing communication quality degradation, such as latency spikes, that occur during inter-cell handovers (HOs). We have been investigating a multi-band redundant vehicle-to-network-to-vehicle (V2N2V) communication scheme as a potential solution to mitigate the impact of such degradation during inter-cell HOs in mobile-network-based inter-vehicle communications. This paper presents a field trial of the proposed multi-band redundant V2N2V scheme under single mobile network operator (MNO), assuming its application to bidirectional real-time control message exchange in automated Bus Rapid Transit (BRT) platooning. Experimental results conducted over a commercial 5G network demonstrate that the proposed scheme can suppress latency spikes during inter-cell HOs. These findings indicate that multi-band redundant V2N2V is a promising approach for achieving stable low latency in V2N2V communication.
Industrial three-phase induction motors are critical to manufacturing operations, yet their failure can cause costly downtime and maintenance overhead. This work introduces a machine learning framework tailored for fault diagnosis and severity assessment in such motors, using synthetic data that replicates six realistic electromagnetic and mechanical fault conditions. The system employs synchronized vibration and current signals, and evaluates several classifiers—including LSTM, GRU, TCN, RNN, and Random Forest—under varying industrial noise profiles such as phase jitter, frequency drift, and transient spikes. Experimental results demonstrate that time-series models consistently outperform classical approaches in noisy environments. For deployment on resource-constrained edge devices, structured pruning and quantization are applied to reduce model size and latency. Beyond motor fault classification, the framework also detects and visualizes approximately spherical clusters to comprehend fault severity levels. The proposed system enables scalable fault monitoring and low-latency predictive maintenance optimized for embedded industrial environments.
Adaptive aquaculture digital twins face cybersecurity challenges due to real-time synchronization, limited device resources, and heterogeneous networks. This paper introduces a multi-layered security framework combining ChaCha20-Poly1305 encryption on ESP32 edge nodes, secure LoRa transmission, and MQTT over TLS 1.3 for cloud communication. A supervisory digital twin monitors integrity and detects anomalies to strengthen resilience against cyber-physical threats. The framework is quantitatively assessed through a normalized security score (Stotal) based on NIST SP 800-53 and ENISA guidelines, achieving 9.0/10 across all layers, while BAN Logic verification ensures mutual authentication and replay protection. Results demonstrate that robust multilayer security can coexist with computational efficiency, enabling scalable and secure Agriculture 5.0 systems.
Helping visually impaired users navigate an athletic track for running is a challenging task that requires a highly reliable, low-latency, real-time navigation system. In this study, we propose a vision-based assisted navigation framework that integrates a camera, a tactile feedback belt, and a computing unit (either an external computer or an on-board embedded device) to detect the running lane and provide real-time directional guidance along the track. Our preliminary experiments relied on a direct Wi-Fi connection between the sensing and processing units. However, the system exhibited severe latency issues and bandwidth limitations, which prevented it from operating effectively in real-time conditions. To overcome these challenges, we propose leveraging vehicular cloud computing as a communication and processing backbone. This approach is expected to significantly reduce latency, improve scalability, and enable more robust real-time navigation assistance for visually impaired runners.
In this paper, we conduct an energy consumption analysis of a broadcast access mechanism over IEEE 802.11 networks. By assuming non-saturated conditions, we provide a statistical model of the energy consumed during one cycle period, defined as the duration between two successful broadcast transmissions. The model includes a characterization of the medium access transmission mechanism. Our results show that the broadcast protocol is a good candidate to be included in future amendments of the IEEE 802.11 standard.
Bluetooth Low Energy and IEEE 802.15.4 are commonly used wireless communication protocols in IoT applications. Both operate in the 2.4 GHz ISM band, where their coexistence can lead to interference. We study the bit error rate (BER) under controlled timed interferences within the physical payload, varying interference power and frequency offsets. Measurements with a co-located software-defined-radio antenna are captured and shown to validate the experimental setup. IEEE 802.15.4 proves more robust to BLE interference than vice versa, consistent with results in literature. Interference at lower relative frequencies is found to have a greater impact on BER. Controlled interference causes persistent bit errors after it ends, suggesting receiver desynchronisation.
Reconfigurable Intelligent Surfaces (RIS) have emerged as a key architectural element in 6G wireless systems, enabling programmable control over electromagnetic (EM) propagation. However, prevailing RIS models are limited in two critical ways: (i) they assume deterministic or Gaussian signal environments, neglecting the inherent spatial randomness of RIS deployments and user locations; and (ii) they treat EM signals as scalars, overlooking the intrinsic multigrade (e.g., scalar, vector, bivector) structure of physical fields. In this paper, we introduce CRISP, a novel Clifford Wiener-Ito Poisson chaos framework for RIS modeling under spatial randomness. By combining stochastic geometry (via Poisson point process), multivector Clifford algebra, and Wiener-It o chaos expansions, we develop a grade-separated stochastic representation of RIS channels. This formulation enables optimal RIS beamforming via per-grade chaos projection, yielding closed-form expressions for grade-wise residual power. Extensive simulations show that CRISP achieves 10-15 dB lower residual power compared to scalar RIS baselines under Poisson-deployed scattering. Our framework provides a new mathematical foundation for RIS optimization in random and semantic wireless environments, and opens new directions in multigrade-aware, chaos-driven RIS control for 6G and beyond.
Tc-CBTC (Train-centric Communication-Based Train Control) system relies on critical communications between trains to ensure the safety and efficiency of operations. However, increasing autonomy of these systems introduces new security challenges related to communication reliability. In this paper, we propose an embedded trust management system for communications in Tc-CBTC system. The proposed solution evaluates the reliability of trains by combining received messages analysis with the long-term trust scores. It uses machine learning techniques (embedding, CNN, LSTM) to analyze behavioral patterns and detect anomalies. Trsut is modeled using beta distribution with dynamic updates based on observed reliability. Decisions making simultaneously consider the reliability of the message and the trust score of the transmitting train. The simulation results demonstrate the high performance of our proposal.
