AIU3 – Applications, IoT, Use cases
Thursday, 4 June 2026, 16:30-18:00, room Sala 12 (1st floor)
Session Chair: Marc Emmelmann (Fraunhofer FOKUS, DE)
Shielding the Edge: A Practical Architecture for CRA-Compliant and Physically-Robust Ambient Intelligence
Vicenç Porras (Universitat Politecnica de Catalunya (UPC), Spain); Julia Igual (i2CAT Foundation, Spain); Marisa Catalan (Universitat Politècnica de Catalunya – BarcelonaTech (UPC), Spain); David Sarabia-Jácome (i2CAT Foundation, Spain)
The emergence of new cybersecurity regulations, particularly the European Cyber Resilience Act (CRA), establishes mandatory requirements for products with digital elements in the Internet of Things (IoT) and Ambient Intelligence (AmI) ecosystems. While current IoT devices often adequately mitigate network-based attacks, they remain critically vulnerable to physical attack vectors that can lead to device malfunction and the theft of Intellectual Property (IP), such as proprietary AI models or sensitive data. This paper addresses this gap by presenting a comprehensive end-to-end security architecture developed in the scope of the European CUSTODES project where we propose a hardware-rooted chain of trust that unifies strict network authentication and encryption with robust device-level physical hardening. The architecture is validated through a real-world use case deploying AmI for Smart Buildings. The proposed solution demonstrates full regulatory compliance and provides effective protection against a comprehensive spectrum of threats proving that robust security can be achieved on Commercial Off The Shelf (COTS) low-cost and resource-constrained IoT hardware, such as the recently emerged Raspberry Pi Pico 2W (RP2350) and a Raspberry Pi 4, without relying on expensive industrial-grade platforms. Finally, this paper serves also as a comprehensive guide for AI IoT practitioners and industry developers trying to understand the implications of the new cybersecurity regulations for their current deployments.
PDU Set-Aware Delay-Reliability-Optimal Downlink Scheduling for Extended Reality Applications in 6G
Moyukh Laha and Gabriel-Miro Muntean (Dublin City University, Ireland)
Extended Reality (XR) is poised to become a flagship application in 6G networks, requiring the joint support of enhanced mobile broadband (eMBB) and ultra-reliable low-latency communications (URLLC)-a service class beyond the capabilities of current 5G systems. To meet these demands, 3GPP Release 18 introduces a Protocol Data Unit (PDU) Set-based quality of service (QoS) framework, necessitating XR scheduling schemes that are both PDU Set-aware and capable of satisfying stringent QoS requirements. This paper presents a downlink scheduling framework tailored for XR services in 6G, incorporating a novel mechanism for tracking PDU Set delay. Employing this framework, we formulate XR scheduling as an optimization problem that enforces PDU Set-specific QoS constraints and solve it optimally to maximize the number of XR users meeting their service requirements. Simulation results demonstrate substantial performance improvements, achieving 20-30% gains over state-of-the-art delay-reliable schedulers and multifold gains over traditional approaches.
Edge-Cloud Orchestration Leveraging Network Exposure APIs for Sustainable Environmental Monitoring over 6G
Konstantinos Nomikos and Sokratis Barmpounakis (WINGS ICT Solutions, Greece); Panagiotis Demestichas (University of Piraeus, Greece)
Next-generation 6G networks must dramatically improve the sustainability of data-intensive services such as real-time environmental monitoring and surveillance. This paper presents a novel Cross-domain Orchestration Engine (COE) that intelligently manages when and where to process video analytics in an Internet-of-Things (IoT) camera system. In contrast to traditional always-on cloud analytics, our COE leverages an event-driven approach and integrates open network exposure APIs (CAMARA Open Gateway) to dynamically allocate network resources on-demand. We define and evaluate three operational modes: an always-on baseline (continuous cloud offload), an event-triggered mode with static offload logic, and a dynamic orchestration mode using our COE metric. The COE-metric driven mode employs content entropy, AI model confidence, and event criticality to decide between local on-camera processing and selective cloud offload, while requesting guaranteed network Quality-on-Demand when needed. Model-based evaluation (24h workload simulation with representative power/latency assumptions) shows that the proposed orchestrated mode reduces energy consumption while meeting low-latency requirements, offering a sustainable 6G monitoring solution with balanced performance and resource efficiency.
Finite-Blocklength URLLC Scheduling with Two-Slot Transmission
Vahid Rajabi (Fraunhofer-Institut, Germany); Patrick Agostini (TU Berlin, Germany); Ehsan Tohidi (Fraunhofer HHI, Germany); Slawomir Stanczak (Technische Universität Berlin & Fraunhofer Heinrich Hertz Institute, Germany)
This paper studies ultra-reliable low-latency com- munications (URLLC) scheduling in the finite blocklength regime. We develop a packet-level framework in which each packet is associated with a finite set of feasible transmis- sion modes, including single-shot transmission, hybrid auto- matic repeat request (HARQ) with Chase combining (CC), and packet segmentation. We formulate a multi user equipment (UE) scheduling problem that maximizes the expected delivered payload per consumed resource element (RE) under latency, reli- ability, and multi-slot resource coupling constraints. To solve this problem, we design a dynamic queue-based weighting mechanism inspired by Lyapunov methods, where backlog and deadline ur- gency adapt the scheduling priorities. Low-complexity algorithms based on linear relaxation (LR) and sequential greedy selection are proposed. Simulation results show significant packet delivery ratio (PDR) improvement over round-robin (RR), proportional fair (PF), and weighted-sum baselines while maintaining fairness, with throughput within about 6% of a no-latency benchmark.
PWE-Assisted 3D Spherical Splatting for Object Composition Visualization Using Detection Transformers
Anastasios T. Sotiropoulos (Aristotle University of Thessaloniki, Greece); Stavros Tsimpoukis (University of Ioannina and AUTH, Greece); Dimitrios Tyrovolas (Aristotle University of Thessaloniki, Greece & University of Patras, Greece); Antonios Lalas (Centre for Research and Technology – Hellas (CERTH), Greece); Sotiris Ioannidis (Technical University of Crete, Greece); Alexandros Ioannis Papadopoulos (University of Ioannina, Greece & Information Technologies Institute, Greece); Panagiotis D. Diamantoulakis and George K. Karagiannidis (Aristotle University of Thessaloniki, Greece); Christos Liaskos (University of Ioannina, Greece & Foundation of Research and Technology Hellas, Greece)
The pursuit of immersive and structurally aware multimedia experiences has intensified interest in sensing modalities that reconstruct objects beyond the limits of visible light. Conventional optical pipelines degrade under occlusion or low illumination, motivating the use of radio-frequency (RF) sensing, whose electromagnetic waves penetrate materials and encode both geometric and compositional information. Yet, uncontrolled multipath propagation restricts reconstruction accuracy. Recent advances in Programmable Wireless Environments (PWEs) mitigate this limitation by enabling software-defined manipulation of propagation through Software Defined Metasurfaces (SDMs), thereby providing controllable illumination diversity. Building on this capability, this work introduces a PWE-driven RF framework for three-dimensional object reconstruction using material-aware spherical primitives. The proposed approach combines SDM-enabled field synthesis with a Detection Transformer (DETR) that infers spatial and material parameters directly from extracted RF features. Simulation results confirm the framework’s ability to approximate object geometries and classify material composition with an overall accuracy of 79.35%, marking an initial step toward programmable and physically grounded RF-based 3D object composition visualization.
