Artificial intelligence is becoming a key enabler for automated radio spectrum monitoring, but progress is still limited by the scarcity of large, standardized datasets derived from real RF measurements. This paper presents SpectrumAI, a large-scale image-based RF corpus and a reproducible methodology for transforming wideband in-phase and quadrature captures into annotated spectrogram representations for machine learning. The proposed pipeline combines scalable signal processing, consistent time–frequency parameterization, and structured labeling to support robust training and fair benchmarking across heterogeneous signal classes. The resulting corpus contains more than 160,000 spectrogram images extracted from real-world measurements, spanning multiple cellular, radar, and communication emissions, while the experimental evaluation in this paper relies on a curated and fully validated benchmark subset of 5,300 images. To validate the practical utility of the dataset, we train and evaluate a YOLO11 model on multiple dataset configurations, including challenging split conditions designed to test generalization. Experimental results confirm that SpectrumAI enables reliable spectrogram-based RF signal detection and localization, while highlighting the central role of data diversity and standardized generation procedures in achieving reproducible performance. Overall, SpectrumAI provides a concrete foundation for advancing AI-driven spectrum intelligence research.

Abstract Migration is a widespread phenomenon across taxa, yet the ecological mechanisms underlying its evolution and maintenance, particularly whether migratory behaviors are primarily driven by access to spatially restricted breeding sites or by seasonal tracking of trophic resources, remain poorly documented outside birds and large mammals. Despite increasing evidence that reptiles perform seasonal migrations, the ecological mechanisms underlying these movements have rarely been formally tested. The critically endangered Galápagos pink land iguana ( Conolophus marthae ), endemic to Wolf Volcano, Isabela Island, exhibits partial migration along a steep altitudinal gradient, providing an opportunity to disentangle the relative roles of breeding-site availability, trophic resource dynamics, and thermoregulatory conditions as drivers of migration. We used GPS tracking data from 22 individuals (7 males, 15 females) monitored between 2019 and 2023, combined with high-resolution spatio-temporal models of vegetation productivity and air temperature across the species’ altitudinal range, to characterize population-level movement patterns and evaluate competing hypotheses explaining the evolution of this migratory behavior. Movement models revealed a clear pattern of partial migration: 16 out of 22 tracked individuals performed seasonal altitudinal movements between a restricted high-elevation mating area and a larger dispersal area at lower elevation, with males reaching the mating area approximately 48 days earlier than females. The dispersal area remained consistently more productive than the mating area throughout the year, rejecting the prediction that individuals should track the shifting trophic resource peaks. Instead, the mating season coincided with the local productivity peak within the mating area, whereas temperature differences between areas were small ( ca. 2°C) and did not explain migration timing. These results support a site-dependent hypothesis of partial migration over a resource-tracking hypothesis, indicating that access to spatially restricted breeding sites is the primary driver of migration in this species, with local trophic resource dynamics fine-tuning reproductive timing. Providing empirical evidence for the ecological mechanisms underlying migration in a large terrestrial reptile, our results extend site-dependent theories of migration beyond birds and mammals and identify breeding-site availability as a key ecological driver of migratory behaviors across taxa.

Instant messaging applications are an integral part of everyday life, facilitating communication in various settings, including work and healthcare. This systematic review aims to analyze the healthcare contexts in which these applications are most commonly used (i.e., work organization, education, communication among healthcare personnel, communication with patients), with a particular focus on privacy issues raised by the handling of sensitive data on these platforms and how, or if, these issues are addressed. Following a systematic literature review process conducted according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, the articles were examined with the goal of extracting the aforementioned information. The results reveal that the majority of studies (82.6%) adopt commercial solutions to improve communication in medical field, with most of them using WhatsApp as the primary communication channel. Although the studies acknowledge the privacy risks introduced by these applications, the description of how such risks are mitigated or addressed is rarely reported in detail; what is consistently highlighted instead is the practical lack of clarity in existing regulations about the pratical applicability.

High-fidelity in-phase and quadrature (I/Q) signal traces are critical for a variety of wireless network applications, including spectrum monitoring, interference detection and mitigation, radio-frequency (RF) fingerprinting (RFFP), smart jamming detection, anomaly identification, and modulation classification. However, the number and scope of publicly available I/Q datasets are currently limited, as most datasets are either restricted to a single frequency or wireless technology, or collected in controlled laboratory environments. This paper introduces ITALYSIG, a comprehensive, high-definition, open-source I/Q database of diverse real-world radio signals, including cellular, radar, and other wireless technologies. ITALYSIG provides I/Q captures with up to 100 MHz bandwidth, collected across diverse urban and rural environments in Italy. The I/Q signals are stored in multiple formats, including raw binary files, standardized VITA Radio Transport (VRT, VITA-49), and visual formats such as Portable Network Graphics (PNG) and JavaScript Object Notation (JSON) for broader applicability. The data-acquisition setup is based on a CRFS RFeye SenS Portable recorder at the front end, which enables automatic long-term I/Q recordings on the order of hours, multi-terabyte storage, and real-time signal processing. In addition to releasing the dataset, this article provides a comprehensive overview and qualitative comparison of state-of-the-art datasets in terms of measurement setup, data format, wireless technology, frequencies, and time duration. Furthermore, this article provides a tutorial on the end-to-end measurement setup for automatic I/Q acquisition, multi-format export, and back-end storage, as well as real-time analysis via the DeepView software. Finally, we provide guidelines for integrating I/Q traces into deep learning pipelines and highlight specific use cases of the dataset.

Agentic AI is emerging as a promising paradigm for network management, enabling high-level intent processing, automated decision making, and closed-loop control. However, the current landscape is fragmented: proposed solutions are often evaluated in ad hoc settings, with limited reproducibility and no common basis for systematic comparison. This lack of benchmarking methodology makes it difficult to assess the actual benefits, limitations, and operational trade-offs of different agentic approaches. This paper presents a playground for benchmarking agentic AI in network management. Rather than proposing a single best-performing agent, the goal is to provide a controlled and extensible environment in which heterogeneous agentic solutions can be deployed, observed, and compared under common network management tasks. The playground combines a programmable multi-node network topology, a transaction-oriented control workflow, structured agent-to-network interfaces, explicit network state representation, and built-in validation and rollback mechanisms. Its design enforces a clear separation between high-level agent reasoning and deterministic execution, thus enabling safer and more auditable experimentation. The proposed framework is instantiated over a network management scenario based on Segment Routing over IPv6 (SRv6), where agentic solutions interact with the infrastructure through declarative messages instead of arbitrary low-level commands. This design supports benchmarking along multiple dimensions, including task success, convergence behavior, robustness to failures, recovery capability, safety of issued actions, and auditability of the control process. By providing a reproducible and observable experimentation environment, the proposed playground lays the foundation for a systematic evaluation methodology for agentic AI in network management.

Spectrum monitoring has become essential to ensure the reliability of wireless communications. Spectrograms provide a visual time-frequency representation of the radio signal, which analyzed with deep learning tools of the visual world, enables the identification of coexistence problems, interference and critical propagation conditions. In this paper, we propose a spectrum analysis system able to detect 5G control signals. Our methodology is based on the acquisition of 16 -bit IQ signals on extended time scales, transformed and analyzed as images. The paper discusses issues related to dataset creation for model training, as well as the possibility of using pre-trained models through transfer learning and few-shot learning to detect signals in real environments. To this end, a proof of concept of the proposed methodology was conducted, based on a dataset created from real-world acquisitions used to train a YOLOv8x model. The results show that the methodology is very promising, but highlight the need for significantly more examples to train the model.

Recent advances in network switch designs have enabled machine learning inference directly within the switch at line speed. However, hardware constraints limit switches capabilities of tracking stateful features essential for accurate inference, as the demand for these features grows rapidly with line rates. To address this, we propose DIDA, a distributed in-network machine learning approach. In DIDA, feature extraction occurs at the host, features are transmitted via in-band telemetry, and inference is performed on the switches. In this paper, we evaluate the effectiveness and efficiency of this architecture. We examine its impact on network bandwidth, CPU and memory usage at the host, and its robustness across different feature sets and deep neural network classifications.

Federated testbeds enable collaborative research by providing access to diverse resources, including computing power, storage, and specialized hardware like GPUs, programmable switches and smart Network Interface Cards (NICs). Efficiently sharing these resources across federated institutions is challenging, particularly when resources are scarce and costly. GPUs are crucial for AI and machine learning research, but their high demand and expense make efficient management essential. Similarly, advanced experimentation on programmable data plane requires very expensive programmable switches (e.g., based on P4) and smart NICs. This paper introduces SHARY (SHaring Any Resource made easY), a dynamic reservation system that simplifies resource booking and management in federated environments. We show that SHARY can be adopted for heterogenous resources, thanks to an adaptation layer tailored for the specific resource considered. Indeed, it can be integrated with FIGO (Federated Infrastructure for GPU Orchestration), which enhances GPU availability through a demand-driven sharing model. By enabling real-time resource sharing and a flexible booking system, FIGO improves access to GPUs, reduces costs, and accelerates research progress. SHARY can be also integrated with SUP4RNET platform to reserve the access of P4 switches.

Satellite systems are increasingly targeted by cyber threats, yet training platforms that reproduce realistic space-ground conditions remain limited. OpenSatRange is a domain-specific cyber range designed to support satellite security training through integrated simulation, emulation, and monitoring capabilities. It combines accurate network simulation (NS-3) for orbital and link modeling, low-level emulation of communication links (OpenSAND), and full-stack scenario deployment using containerized sandboxes orchestrated via KYPO and SDN. Unlike generic cyber ranges, OpenSatRange supports both LEO and GEO constellations and enables detailed observability of trainee actions through embedded telemetry, facilitating real-time instructor feedback and post-exercise analysis. We describe the system architecture, deployment workflow, and telemetry design, and present illustrative training scenarios such as broadcast hijacking and insecure ground control that demonstrate the platform's flexibility and educational impact. The platform enables training that goes beyond binary outcomes, offering visibility into the reasoning processes that lead to success or failure, crucial for effective cybersecurity education in the space domain.

Extensible In-band Processing (EIP) was proposed to support advanced in-network operations through flexible, programmable metadata carried in packet headers. This work integrates EIP into the IETF IOAM architecture by encapsulating it as a new IOAM Data-Field-Type, leveraging the IOAM processing model while preserving EIP extensibility. The approach eases deployment for IOAM operators and unifies telemetry and control-plane signaling. The paper presents design alternatives and compatibility considerations, and highlights the benefits and limitations of aligning EIP with the IOAM processing pipeline.

Accurate localization is a key enabler for advanced autonomous applications in UAVs and Internet of Things (IoT) networks, particularly in outdoor scenarios where GPS may be unavailable, unreliable, or insufficiently precise. Ultra-Wideband (UWB) technology offers high-accuracy ranging capabilities and has shown strong performance in indoor environments, but its deployment in large-scale outdoor networks remains underexplored. This work investigates the presence of range-dependent bias in long-distance UWB measurements and evaluates its impact on multilateration-based localization systems for UAVs. We present a field measurement campaign with specifically designed UWB testing boards over distances up to 1.2 km, revealing that all tested modules exhibit a growing positive bias, up to 20 cm at the maximum range. To assess how such bias affects positioning, we implement a simulation framework modeling a UWB localization network. We analyze the effect of varying anchor spacing and compare ideal measurements to biased ones using the measured trend. Our results show that geometric anchor deployment can mitigate some of the bias effects, and that awareness of systematic errors is crucial for the design of scalable, accurate outdoor localization networks.

Distributed training of artificial intelligence models, such as Large Language Models (LLMs), generates highly structured and intense traffic patterns between GPUs, with synchronous and repetitive flows that can easily cause congestion and bottlenecks in data center networks. In this context, currently adopted protocols, such as RoCEv2, show significant limitations in the presence of bursty traffic and low entropy, compromising overall system efficiency. Segment Routing over IPv6 (SRv6) offers a programmable mechanism to steer AI workload traffic along explicitly chosen paths, enabling precise and congestion-aware routing under dynamic conditions. Lightweight monitoring modules can detect congestion conditions in real time and report them to the orchestrator or NICs, enabling dynamic rerouting decisions without requiring control-plane signaling or state in the fabric. SRv6 micro-segment (uSID) encoding allows the NIC to steer traffic along alternate, congestion-free paths simply by updating the IPv6 destination address, preserving RoCEv2 semantics while ensuring rapid adaptability. This work provides a practical implementation and experimental validation of the recent IETF Internet-Draft “SRv6 for Deterministic Path Placement in AI Backends”, demonstrating its feasibility and performance benefits in RoCEv2-based infrastructures. The results highlight the potential of SRv6 as a practical and vendor-agnostic solution to enhance networking efficiency in modern AI datacenters.

Electronic Patient-Reported Outcome Measures (ePROMs) are widely used in telemonitoring for efficient patient status assessment, without the need of clinician intervention. However, the quality of collected data is often compromised by issues such as patient comprehension, response fatigue, and varying levels of digital proficiency. We’re looking at ways to overcome these challenges by using crowdsourcing techniques to evaluate the quality and reliability of patient responses. Our idea is to enhance PROMs by adding elements inspired by crowdsourcing, such as asking repeated or counterfactual questions, gauging self-reported confidence, and checking internal response consistency. This is all about improving how we understand and assess patient feedback without changing the original questionnaire’s structure. To show how this can work in practice, we’ve designed a proposal using the SNOT-22 questionnaire, which is used in otolaryngology to manage chronic upper airway diseases. Our plan involves adding extra questions and using metadata analysis to indirectly but effectively enhance response quality.

The monitoring of wildlife through advanced tracking technologies is crucial for ecological research, conservation efforts, and the study of animal behavior. However, most existing tracking devices are primarily designed for positional data collection, with limited integration of biologically relevant sensors. This work presents the design, development, and evaluation of a low-power, multisensor tracking device tailored for long-term wildlife monitoring. Our device integrates a comprehensive set of sensors and novel methodologies to acquire additional information about animals, including monitoring heart rate and respiration in free-ranging individuals. We optimize energy consumption through an advanced power management system, leveraging energy harvesting to ensure long-term operational sustainability. We validate the system with extensive laboratory tests on energy efficiency and data accuracy. This research paves the way for next-generation wildlife tracking devices capable of supporting more comprehensive studies in conservation biology and environmental monitoring.

The healthcare industry is witnessing a rapid rise in the adoption of wearable and implantable medical devices, including advanced electrochemical sensors and other smart diagnostic technologies. These devices are increasingly used to enable real-time monitoring of physiological parameters, allowing for faster diagnosis and more personalized care plans. Their growing presence reflects a broader shift toward smart connected healthcare systems aimed at delivering immediate and actionable insights to both patients and medical professionals. At the same time, the healthcare industry is increasingly targeted by cyberattacks, primarily due to the high value of medical information; in addition, the growing integration of ICT technologies into medical devices has introduced new vulnerabilities that were previously absent in this sector. To mitigate these risks, new international guidelines advocate the adoption of best practices for secure software development, emphasizing a security-by-design approach in the design and implementation of such devices. However, the vast and fragmented nature of the information required to effectively support these development processes poses a challenge for the numerous stakeholders involved. In this paper, we demonstrate how key features of the Malware Information Sharing Platform (MISP) can be leveraged to systematically collect and structure vulnerability-related information for medical devices. We propose tailored structures, objects, and taxonomies specific to medical devices, facilitating a standardized data representation that enhances the security-by-design development of these devices.

The Common Vulnerability Scoring System (CVSS) is the de facto standard for assessing the severity of cybersecurity vulnerabilities. Yet, in practice, it is often misused. Most stakeholders rely solely on the so called CVSS "Base Score", a measure of technical severity, ignoring the contextual metrics introduced in the 4.0 version: "Threat" and "Environmental". This narrow use results in poor prioritization, especially in safety-critical domains like healthcare. In this study, we perform the first large-scale application of CVSS 4.0's full Base, Threat, Environmental (BTE) scoring to a comprehensive dataset of medical device vulnerabilities. We show that the Threat group can be partially automated using structured data sources, while meaningful environmental profiles (e.g., home vs hospital care) allow semi-automatic compilation of the Environmental metrics. Our results confirm that BTE scoring significantly changes vulnerability prioritization, yielding a representation of risk that is both more accurate and more actionable, especially when safety is involved.