Thursday, 4 June 2026, 12:30-13:00, Sala de Exposiciones 1
Session Chair: tbd
P-2.1 A Dynamically Beam Steerable Metasurface Antenna with Sidelobe Suppression
Enez Furkan Cihan (ASELSAN Inc., Turkey)
A dynamically beam-steerable metasurface antenna based on a substrate integrated waveguide is presented. Metasurface elements are loaded with PIN diodes to dynamically control the beam steering toward different directions. A modified sidelobe suppression method is also incorporated alongside beam steering and evaluated using analytical array factor calculations and full-wave simulations in ANSYS HFSS. The proposed antenna achieves up to 6 dB sidelobe reduction with minimal main-beam degradation while steering the beam from −25° to +30° in the Ku band using only 24 elements. The compact SIW-based architecture offers low-cost and low-complexity solution for next-generation sensing and communication applications.
P-2.2 Design and Experimental Validation of a 1-Bit PIN Diode Based 6.4 GHz RIS Tile
Luís Outeiro, Francisco G. Vilarinho and Sofia I. Inacio (INESC TEC, Portugal); Mohamed Elsaid Ghatas (INESC TEC, Campus Da FEUP, Portugal & University of Porto, Portugal); Nuno Paulino (INESC TEC, Portugal & University of Porto, Portugal); Luis M. Pessoa (INESC TEC & Faculty of Engineering, University of Porto, Portugal)
This work presents the design, implementation, and experimental validation of a Reconfigurable Intelligent Surface (RIS) aimed at Wi-Fi 6 applications. An 8×8 modular tile architecture is introduced for physical scalability and decentralized control. The hardware integrates a control board with USB-C Power Delivery (PD), allowing multiple tiles to operate from a single source. Beamforming computation is executed on an embedded ESP32 using an MQTT-based framework for remote distributed control. Validation through waveguide characterization and anechoic chamber measurements showed that although the PIN-diode unit cells were designed for 6.5 GHz, component tolerances shifted optimal operation to 6.38 GHz, where effective beam steering at 0°, ±20°, and ±40° was demonstrated.
P-2.3 Slice-Aware 5G Backhaul for End-to-End Network Slicing, Isolation, and QoS Assurance
Xhulio Limani (University of Antwerp, Belgium & Imec, The Netherlands); Andreas Gavrielides (University of Antwerp, Belgium & IMEC, Belgium); Miguel Camelo Botero (University of Antwerp – imec, Belgium); Johann Marquez-Barja (University of Antwerpen & IMEC, Belgium); Nina Slamnik-Krijestorac (University of Antwerp-IMEC, Belgium)
Fifth-generation (5G) network slicing enables multiple logical networks to share the same infrastructure, each designed to support distinct Service Level Agreements (SLAs). While network slicing is well supported in the Radio Access Network (RAN) and 5G Core (5GC), the Transport Network (TN) backhaul is often treated as a slice-agnostic forwarding substrate. As a result, resource contention on shared links can significantly degrade network performance, such as latency and throughput, breaking end-to-end SLAs. We aim to address this gap by making the TN slice-aware through a lightweight framework that translates slice requirements into concrete backhaul policies. The framework (i) associates each slice with TN identifiers at domain boundaries, (ii) turns slice requirements into per-slice forwarding and scheduling policies (e.g., priority and queue selection, bandwidth bounds, and optional congestion signals), and (iii) exports per-slice measurements for monitoring and closed-loop control. We deploy a proof of concept on a programmable switch and evaluate it on a real-life 5G testbed. Our results show that, under link saturation, slices designed for mission-critical services preserve their SLAs, while best-effort traffic degrades predictably.
P-2.4 Enricher-Assisted Sensing-Aided Beam Prediction for mmWave UAV Communications in iSEE-6G: A JCCSP Architecture for CCAM and PPDR
Andreas Gavrielides (University of Antwerp, Belgium & IMEC, Belgium); Rreze Halili (University of Antwerp – imec, Belgium); Johann Marquez-Barja (University of Antwerpen & IMEC, Belgium); Nina Slamnik-Krijestorac (University of Antwerp-IMEC, Belgium)
Future 6G systems will support dynamic aerial and ground services where communication, sensing, and edge intelligence are tightly integrated. In this context, the SNS JU iSEE-6G project advances a Joint Communications, Computation, Sensing and Power transfer (JCCSP) framework for UAV-centric 6G environments, including physical-layer beamforming, MEC-supported positioning, and emergency-response applications. This paper proposes an architecture that links sensing-assisted beam prediction for mmWave UAV communications with the Enricher, an edge intelligence component that fuses network, sensing, and application telemetry for closed-loop adaptation. The approach exploits multimodal observations, including radio measurements, positioning, and environment-aware context, to anticipate beam directions and trigger proactive edge-assisted actions. Recent studies show that multimodal sensing improves beam prediction robustness in mobile mmWave and ISAC scenarios, including UAV and vehicular settings. Building on these advances, we position the Enricher as a cross-layer control element that translates beam prediction outputs into service-aware orchestration decisions for two iSEE-6G verticals: connected and cooperative automated mobility (CCAM) and public protection and disaster relief (PPDR). In CCAM, the workflow supports resilient UAV-assisted roadway awareness and low-latency traffic supervision. In PPDR, it supports mission-critical connectivity through predictive beam maintenance and edge-based adaptation under rapidly changing conditions.
P-2.5 Poster – SRv6 Vs ULCL in Multi-UPF 5G/6G Networks: An Open-Source Edge Computing Evaluation
Bilal Ghani and Hakima Chaouchi (Télécom SudParis, France)
Uplink Classifier (ULCL)-based traffic steering at the Intermediate-User Plane Function (I-UPF) relies on GPRS Tunneling Protocol for User Plane (GTP-U) inspection and N9 tunneling, increasing system complexity and limiting cost-efficient 6G deployments. This work compares ULCL with an SRv6-based forwarding approach that removes GTP-U encapsulation on the N9 interface without modifying the access network or control plane. Using an open-source testbed with free5GC, UPG-VPP, and UERANSIM, we evaluate latency, throughput, and CPU utilization. Results show that SRv6 can effectively replace ULCL in multi-UPF 5G/6G networks without redesigning the user-plane architecture.
P-2.6 An Enhanced Hammerstein Model for High-Fidelity Digital Pre-Distortion of OFDM Waveforms
Jinfei Wang, Yupeng Zheng, Yi Ma and Rahim Tafazolli (University of Surrey, United Kingdom (Great Britain)); Irshaad Fatadin (National Physical Laboratory, United Kingdom (Great Britain)); Anthony Page (Keysight Technologies, United Kingdom (Great Britain))
The design of high-fidelity digital pre-distortion (DPD) for orthogonal frequency-division multiplexing (OFDM) waveforms is challenged by high training complexity. To address this challenge, we propose an enhanced Hammerstein (e-Hammerstein) DPD model that replaces the non-linearity compensation block of the conventional Hammerstein structure with peak-to-average power ratio (PAPR) reduction algorithms. With this reformulation, the DPD training process is simplified to the estimation of a linear filter, resulting in substantially reduced training complexity. Moreover, by suppressing high-amplitude signal components prior to amplification, the proposed e-Hammerstein model achieves improved linearization performance. Measurement results demonstrate that the e-Hammerstein model consistently outperforms the generalized memory polynomial (GMP) model across multiple DPD metrics, achieving approximately 3 dB improvement in error vector magnitude (EVM) for Wi-Fi 7 waveforms and around 2 dB improvement in adjacent channel leakage ratio (ACLR) for 5G New Radio waveforms.
P-2.7 Carbon Footprint-Aware Gateway Placement in LoRa Multi-Hop Networks: An Input Importance Analysis for Green Communications
Francisco-Jose Alvarado-Alcon (Universidad Politécnica de Cartagena, Spain); Rafael Asorey-Cacheda, Joan Garcia-Haro and Antonio-Javier Garcia-Sanchez (Technical University of Cartagena, Spain)
Green communications require planning methods that explicitly optimize environmental impact, not only performance or energy. We study what information is truly necessary for a multilayer perceptron (MLP) that recommends gateway positions to minimize the carbon footprint (CF) of LoRa-based multi-hop Internet of Things (IoT) networks. Using the CF model and ILP-labeled dataset as in our prior work, we retrain the MLP and evaluate four input encodings: end-device coordinates, traffic-based device weights, spatial sampling regions (SSRs), and a scalar CF estimate of the current network. Through different techniques we show that the CF estimate is the most influential input and acts as a coarse global cue that triggers large gateway relocations, while spatial encodings (coordinates+SSRs) provide the local detail needed for fine placement. We translate these findings into concise design guidelines for CF-aware learning pipelines suitable for scalable and potentially real-time green network planning.
P-2.8 An Intelligent Automated Penetration Testing Method for Telecommunication Networks
Alla Pinchuk, Roman Odarchenko, Oleh Polihenko, Viktor Hnatiuk, Oleksii Assaul and Kateryna Ishchuk (State University "Kyiv Aviation Institute", Ukraine)
Telecommunication networks are complex, large-scale and business-critical, making manual penetration testing slow, costly and hard to perform regularly. Existing automated and AI-based solutions are largely designed for generic IT environments and overlook telecom-specific architectures and protocols. This paper proposes an intelligent automated penetration testing method that integrates formal modelling and attack graphs, LLM-based scenario planning, MARL-based attack-path optimization, and policy-driven orchestration of classical and telecom-specific tools.
P-2.9 Resource-Aware near-Edge Distributed Deployment of Large Language Models
Ahmed Salah Tawfik Ibrahim (CNIT, Italy); Rana Abu Bakar (Scuola Superiore Sant'Anna, Italy & CNIT, Italy); Emilio Paolini (Scuola Superiore Sant'Anna, Italy); Juan Jose Vegas Olmos (NVIDIA, Denmark); Francesco Paolucci and Filippo Cugini (CNIT, Italy)
Large language models are spreading out into many applications, making their proximity to the end-user crucial, which necessitates the development of techniques and/or the design of more powerful hardware to enable such deployments. In this paper, we investigate the capabilities of the newly released Nvidia DGX Spark as an enabler for edge-oriented LLM platforms. By setting up a system consisting of two such devices, where one complements the other based on resource requirements, we show the promising opportunities this device presents. In doing so, we demonstrate that an LLM deployment closer to the edge is feasible with these devices. In addition, we provide a method to orchestrate a dynamic deployment considering the available resources. Through experiments on the Llama 3 8B and 70B-Instruct models, we show that a 70-billion-parameter model can be deployed at full 16-bit floating-point precision, and that the dynamic distributed deployment incurs negligible performance loss compared to its single-device counterpart.
P-2.10 Optimizing 1-Bit ADC Quantization Threshold in Massive SIMO Systems for Enhanced Detection
Rina Takagi and Kenta Umebayashi (Tokyo University of Agriculture and Technology, Japan); Bikshapathi Gouda, Janne Lehtomäki and Antti Tölli (University of Oulu, Finland)
We study quantization-threshold design for 1-bit analog-to-digital converters (ADCs) in uplink single-input multiple-output (SIMO) systems employing higher-order modulation, specifically 16QAM. To further improve the reliability of SIMO communication systems with 1-bit ADCs, this paper proposes adjusting the 1-bit ADC quantization threshold, in contrast to conventional designs that fix it at zero. The proposed method optimizes the 1-bit ADC quantization threshold with an optimal detector that explicitly accounts for received-symbol statistics, including the nonlinear distortion from 1-bit quantization. Numerical results show that the proposed design achieves symbol error rates (SER) as low as 10^, compared to about 10^ with a conventional zero-threshold detector at 20 dB SNR.
P-2.11 Photonic-Enabled RF Carrier Generation and Distribution for Low Phase Noise ISAC Systems
Zichuan Zhou, Zun Htay, Amany Kassem and Kari Clark (University College London, United Kingdom (Great Britain)); Zhixin Liu (University College Londo, United Kingdom (Great Britain)); Izzat Darwazeh (University College London, United Kingdom (Great Britain))
Integrated sensing and communications (ISAC) is a key enabler for future 6G systems, identified to provide new functions, capabilities and services and is expected to facilitate sensing with increased levels of accuracy. In ISAC systems, carrier phase noise plays a crucial role because it fundamentally determines sensing accuracy and data capacity. In addition, low phase noise radio frequency (RF) carriers need to be frequency synchronized among communication and sensing nodes to ensure accurate sensing and low-latency communications. In this work, we propose and experimentally demonstrate a novel optical frequency comb based technique for highly-scalable low phase noise frequency synchronized RF carrier generation and distribution. A wideband optical frequency comb is generated at the DU (distributed unit) side and can be distributed to multiple RUs (radio units) through optical fibre links. The proposed photonic-enabled carrier exhibits four times lower rms jitter (34fs vs. 135.8fs) compared to a conventional electronics carrier at 25GHz. In this proof-of-concept experiment, we show receiver sensitivity improvement of 3dB using the proposed photonic-enabled carrier generation compared to conventional electronic carrier generation for 16QAM-1Gbit/s signals centred at 25GHz, enabled by the lower phase noise carrier.
P-2.12 Implementation of a V2X-Enabled Collective Perception Architecture for Urban Road Safety
Ivan Huerta (i2CAT Foundation, Spain); Sergi Mercadé (Fundació i2CAT, Internet i Innovació Digital a Catalunya, Spain); Jordi Marias-i-Parella (i2CAT Foundation, Spain & Universitat Politècnica de Catalunya, Spain); Marc Codina (i2CAT, Spain); Francisco Paredes and Miguel Fornell (Idneo Technologies, Spain); Francisco Vázquez-Gallego (i2CAT Foundation, Spain)
Modern Advanced Driver-Assistance Systems (ADAS) are inherently limited by the direct line-of-sight (LoS) of onboard sensors. This paper presents a comprehensive framework for Collective Perception (CP) that leverages Vehicle-to-Everything (V2X) communication and late data fusion to extend a vehicle's sensing range. We focus on a specific deployment in an urban environment involving vulnerable road user (VRU) awareness through infrastructure relaying, specifically addressing scenarios where direct vehicle-to-vehicle (V2V) communication is obstructed by static and dynamic urban obstacles. The proposed architecture utilizes standardized Collective Perception Messages (CPM) to share detected object metadata, optimizing communication load while maintaining high perception accuracy. We provide an exhaustive dive into the implementation of the hardware-software stack, the MOTCAM perception stack for infrastructure, and the V2X FlexStack for ETSI-compliant V2X communications and an evaluation of the performance of the proposed architecture in a use case of VRU protection in an urban scenario.
P-2.13 A Method for Early Prediction of Spectrum Utilization Increases Using User Behavior Trends
Yau Hong Leow and Kenta Umebayashi (Tokyo University of Agriculture and Technology, Japan); Sergio Infante (ITIS Software, University of Málaga, Spain); Julia Robles (University of Málaga, Spain); Manuel Diaz and Cristian Martín (University of Malaga, Spain); Su Pyae Sone (Finland)
This paper tackles early prediction of increases in spectrum utilization by leveraging causal information, specifically user arrival that precedes traffic on the air. In contrast to conventional methods that extrapolate from historical utilization data and therefore react only after increases occur, we pursue early increase prediction from a causal perspective based on detected user arrivals. Experimental results from two indoor Wi-Fi environments (laboratories in Japan and Spain) show that user arrival enables advance declaration of DC increases in Japan in 74% of cases, typically by 0-4 minutes (occasionally longer), whereas a historical-only baseline remains reactive. In Spain, incorporating site-specific user behavior trends improves performance, enabling early prediction in 83% of cases.
P-2.14 Towards Emergent Capabilities in 6G
Aleksandr Zavodovski and Sanna Tuomela (University of Oulu, Finland); Abhishek Kumar (University of Jyväskylä, Finland); Jussi Kangasharju (University of Helsinki, Finland); Ari T. Pouttu (Centre for Wireless Communications University of Oulu, Finland)
The evolution toward 6G marks a fundamental transition from a purely performance-driven network to an integrated ecosystem that natively embeds intelligence, delivers context-aware services, and is designed with sustainability as a first-order principle. Existing AI-driven network management approaches largely rely on centralized or hierarchical orchestration, with pre-defined services optimized for well-known objectives. That approach may become insufficient for future 6G ecosystems characterized by extreme heterogeneity, highly dynamic service demands, and sustainability-driven design. This paper proposes a next step for the 6G paradigm: moving from the Service-of-Services vision towards emergent capabilities, in which largely autonomous, goal-driven AI agents operate across the network-compute-service continuum. These agents exhibit learning and coordination capabilities that allow them to negotiate resources and objectives in real time and to dynamically assemble ad hoc services, among other functions. Collectively, such behaviors give rise to what we tentatively term emergent capabilities – transient, context-aware, and goal-oriented constructs that arise dynamically from multi-agent interactions in response to both exogenous and endogenous factors. In this sense, the system can both respond to explicit user demands and proactively anticipate future needs by inferring intent, context, and evolving objectives from observed interactions and environmental signals. To introduce our vision, we present a high-level architecture and outline use cases that demonstrate the viability and practicality of the approach, with a focus on sustainability.
P-2.15 Validating Remote Driving via Field Characterization and Inter-Slice Connectivity
Raul Lozano, Ivan Viciedo, Jorge Roche Peris and David Gomez-Barquero (Universitat Politecnica de Valencia (UPV), Spain); Jaime Jesús Ruiz Alonso and Ignacio Benito Frontelo (Nokia XR Lab, Spain); Carles Navarro Manchón and Matteo Pagin (Keysight Technologies, Denmark)
The proliferation of 5G Stand-Alone Non-Public Networks (NPNs) enables strict Quality of Service (QoS) for remote driving, yet interconnecting isolated NPNs remains a significant architectural challenge. This paper proposes and validates a lightweight Inter-Slice Federation framework that extends local network slice contexts to remote domains via secure VPN tunneling at the user plane. We report on a cross-site field trial connecting a Host NPN at the Nokia office in Madrid to a Visiting NPN at the Universitat Politècnica de València (UPV). To evaluate the system, we employ a complementary two-phase methodology. First, we conduct a rigorous radio characterization of the Visiting NPN (n40 band) using Keysight Nemo and IxChariot to establish the baseline coverage and capacity in a single-user scenario, as well as to analyze the influence of RAN-level QoS policies in a multi-user scenario. Second, we functionally validate the Inter-Slice architecture by deploying an Autonomous Mobile Robot (AMR) in the Visiting NPN controlled via an immersive cockpit in the Host NPN.
P-2.16 Optimal Wavelength and Bandwidth Evaluation for Satellite-to-Underwater Optical Communications
Alessandro Ugolini and Federica Poli (University of Parma, Italy); Stefano Mangione (Università degli Studi di Palermo, Italy); Daniele Croce (University of Palermo, Italy)
Communication between satellites and underwater devices poses significant challenges due to the heterogeneous characteristics of the channel, traversing free space, atmosphere, and underwater. In this paper, we develop a complete channel model for a direct satellite-to-underwater communication link operating in the visible spectrum. We show that the optimal wavelength mostly depends on the water characteristics and demonstrate that several channels could be allocated using the wavelength division multiplexing technique to increase performance.
P-2.17 Towards Sustainable Resilience to Technology-Aided Adversaries in Future Communication Networks
Tharaka Mawanane Hewa and Nisita Weerasinghe (University of Oulu, Finland); Jonghyun Kim (Sejong University, Korea (South)); Mika E Ylianttila (University of Oulu, Finland)
The connectivity is a critical requirement of the society. Communication networks, including 6th Generation (6G) and future networks promise to deliver extensive capabilities with intelligent and automated services while preserving the sustainability. With the ubiquitous deployment architecture with broadened attack surfaces, the security and resilience are important aspects. In this paper, we explore the intriguing technologyaided attacks, how they breach the sustainability objectives, and relevant research directions to achieve resilience in the communication networks. Our work provide insights for refining the current security architecture with preparedness for future attacks and ensure resilience while preserving sustainability goals.
P-2.18 Design and Evaluation of a Proximity-1/LoRa Transceiver for Control-Plane Inter-Satellite Links
Alex Andreo (i2CAT Foundation, Spain); Elena Fernandez-Nino (Technical University of Catalonia (UPC), Spain & Space Communications Research Group, I2CAT Foundation, Spain); Juan A. Fraire (Inria/INSA Lyon & CONICET, National University of Córdoba, France); Adriano Camps (UPC BarcelonaTech, Spain); Joan Adria Ruiz-de-Azua (Universitat Politècnica de Catalunya (UPC), Spain & Space Communications Research Group, i2CAT Foundation, Spain)
The narrow beamwidth of highly directional Radio-Frequency or optical inter-satellite links imposes stringent requirements on the Pointing, Acquisition, and Tracking (PAT) process. This work presents the design, implementation, and experimental evaluation of a Proximity-1/LoRa transceiver conceived as a low-data-rate, wide-beam auxiliary channel for satellite-to-satellite control-plane communications. The proposed stack combines the CCSDS Proximity-1 data-link protocol over a LoRa physical (PHY) layer to exchange coordination messages under relaxed pointing constraints. To the best of the authors' knowledge, this protocol combination has not been experimentally demonstrated for satellite-to-satellite links. Experimental results show that, compared with a standalone LoRa PHY operating on a congested channel, the proposed transceiver improves reliability by dynamically negotiating a working channel, thus providing an auxiliary link to accelerate the PAT process while effectively avoiding interference.
P-2.19 Enabling Vertical Services with a Cross-Domain 6G Architecture and Trial Framework: The AMAZING-6G Approach
Nina Slamnik-Krijestorac (University of Antwerp-IMEC, Belgium); Francois Carrez (University of Surrey, United Kingdom (Great Britain)); Ewout Brandsma and Sarah N Lim Choi Keung (TNO, The Netherlands); Tanya Politi (University of Patras, Greece); Giada Landi and Gabriele Scivoletto (Nextworks, Italy); Daniele Brevi and Edoardo Bonetto (Fondazione LINKS, Italy); Andreas Georgakopoulos (WINGS ICT Solutions, Greece); Henrik Brun (Oslo University Hospital, Norway); Ole Elle (Rikshospitalet, Oslo University Hospital, Norway)
Beyond 5G (B5G) and 6G networks promise advanced performance and flexibility for emerging vertical applications, but the full adoption of advanced technologies and features and their real-world validation remains limited. This paper focuses on the AMAZING-6G architecture for a large-scale trial framework, which aims to empower vertical industries through an integrated compute and networking continuum. This platform delivers key capabilities such as Compute-as-a-Service, Network Slicing-as-a-Service, AI-as-a-Service, and energy-efficient IoT support to improve performance and energy efficiency of various vertical industries, stretching over healthcare, public safety, transport, and energy domains. The paper introduces the system architecture and highlights four representative use cases and their adoption of B5G/6G architectural components.
P-2.20 Techno-Economic Assessment of Converged Mobile-Access/Metro Networks Enabled by Coherent Pluggable Transceivers and IPoDWDM Technology
José Manuel Rivas-Moscoso (Telefónica Innovación Digital, Spain); Alfonso Sánchez-Macián (Universidad Carlos III de Madrid, Spain); Antonio Melgar González (Telefónica Innovación Digital, Spain); Juan Pedro Fernández-Palacios (Telefónica I+D, Spain); David Larrabeiti (Universidad Carlos III de Madrid, Spain); José David Martínez Jiménez (Telefónica, Spain); Marco Quagliotti (Telecom Italia, Italy); Eduardo. Yusta Padilla and Iván de Francesca (Telefonica, Spain)
Recent advances in coherent pluggable transceivers enable IP over DWDM (IPoDWDM) architectures that allow traffic to be injected directly into the optical layer, potentially bypassing intermediate aggregation nodes. In this paper, we present a techno-economic assessment of an IPoDWDM approach applied to a realistic metro-access and metro-regional network topology inspired by operator deployments. The proposed architecture, based on 100 Gb/s and 400 Gb/s coherent pluggable transceivers, is compared against a conventional IP+DWDM baseline. Results show that optical bypass of access aggregation nodes is technically feasible within standard DWDM reach and spectral constraints, while enabling a reduction in total network cost of over 10% and a power consumption decrease exceeding 40%. These findings highlight IPoDWDM as an effective solution for scalable and energy-efficient metro network evolution.
P-2.21 Enhancing Secure Intent-Based Networking with an Agentic AI: The EU Project MARE Approach
Iulisloi Zacarias (Technische Universität Braunschweig, Germany); Marla Grunewald (Technische Univeristät Braunschweig, Germany); Fin Gentzen (Technische Universität Braunschweig, Germany); Xavi Masip-Bruin (Universitat Politècnica de Catalunya (UPC) & Advanced Network Architectures Lab (CRAAX), Spain); Admela Jukan (Technische Universität Braunschweig, Germany)
In the EU project MARE, a novel plane was proposed, and used in combination with intent-based networking (IBN), allowing the operator to focus on what, rather than on how. Recently, large language models (LLMs) have been successfully employed to translate the high-level intents into low-level actions. The open challenge is to understand how IBN can be effectively enhanced with LLMs and the emerging agentic AI for security purposes. Enhancing IBN with an agentic AI paradigm introduces significant challenges that existing solutions do not fully address. This poster proposes an enhanced IBN framework with a strong security focus toward agentic AI. We address the architectural and security requirements for a multi-agent intent-based system (IBS) architecture, including a multi-domain IBN. We propose a hierarchical multi-agent and multi-vendor architecture that can also be applied more broadly in 6G architectures, and beyond the security architecture proposed in MARE. The architecture incorporates an interactive intent-processing pipeline using LLMs and allows the IBS to connect to external security knowledge bases, such as MITRE ATT&CK, MITRE FiGHT, and NIST.
P-2.22 Dynamic Authorization for Knowledge-Base Agents in 6G
Loay Abdelrazek (Ericsson, Sweden); Leyli Karaçay (Ericsson Research, Turkey); Marin Orlić (Ericsson Research, Sweden)
As 6G architectures transition toward decentralized Multi-Agent Systems (MAS), ensuring secure access to shared Knowledge Bases (KB) is critical. Traditional authorization models like RBAC fail to provide the granularity required for autonomous agents interacting with Semantic-based data. This work proposes a hybrid authorization framework that integrates roles and First-Order Logic (FOL) predicates to enforce zero-trust principles at the knowledge-graph level. We eliminate permission inheritance by enforcing authorization at the triple level (Subject-Predicate-Object), ensuring agents only access metadata required for their specific functional lifecycle.
P-2.23 RIS-Integrated Full-Duplex UAV SWIPT Relaying for Next-Generation Non-Terrestrial Networks
Manojkumar B. Kokare, Neha Sharma, Swaminathan R and Sumit Gautam (Indian Institute of Technology Indore, India)
In this work, we investigates a full-duplex (FD) uncrewed aerial vehicle (UAV)-assisted non-terrestrial network (NTN) system integrating reconfigurable intelligent surface (RIS) with random reflection coefficients (RRC) and simultaneous wireless information and power transfer (SWIPT), where the UAV operates as a decode-and-forward (DF) relay enabling joint relay and RIS transmission between a base station and a ground user. Selection combining (SC) is employed at the receiver, and system performance is evaluated in terms of outage probability (OP) and ergodic capacity (EC) over α-μ fading. Closed-form expressions, validated via Monte-Carlo simulations, demonstrate notable performance gains over conventional schemes, highlighting the effectiveness of the proposed framework for emergency communication scenarios.
P-2.24 Preliminary Approach to Voice-Driven Semantic Perception for UAV-Assisted Emergency Systems
Nuno Saavedra (INESC TEC & Universidade Do Porto, Portugal); Pedro Ribeiro (INESC TEC, Portugal & Universidade do Porto, Portugal); André Coelho (INESC TEC, Portugal); Rui Campos (INESC TEC and Faculty of Engineering, University of Porto, Portugal)
Unmanned Aerial Vehicle (UAV)-assisted networks are increasingly foreseen as a promising approach for emergency response, providing rapid, flexible, and resilient communications in environments where terrestrial infrastructure is degraded or unavailable. In such scenarios, voice radio communications remain essential for first responders due to their robustness. However, their unstructured nature limits direct integration with automated UAV-assisted network management. This paper proposes a preliminary version of SIREN, an AI-driven framework that enables voice-driven perception for UAV-assisted networks.
P-2.25 NL-COMM-Sat: Breaking the Direct Device-to-Satellite Communication Barrier via "Aggressive" Non-Orthogonal Transmissions and Non-Linear Processing
Konstantinos Nikitopoulos and Chathura Jayawardena (University of Surrey, United Kingdom (Great Britain))
Direct Device-to-Satellite (D2S) communications, which enable direct satellite connectivity with unmodified user equipment (UE), not only expand global coverage but also reshape the evolution of future access networks. However, D2S links face fundamental challenges due to inherently low signal-to-noise ratios (SNRs) and limited spatial multiplexing gains arising from near line-of-sight propagation, both of which severely constrain achievable spectral efficiency. Despite the lack of spatial multiplexing, this work shows that aggressive non-orthogonal transmissions, where multiple users (e.g., four) transmit concurrently over the same frequency resources, even to a single receive antenna, can unlock substantial capacity gains that remain entirely unexploited by existing systems. Realizing these gains in practice, however, requires receiver architectures that, to the best of our knowledge, have not yet been developed. To this end, we introduce NL-COMM-Sat, an efficient and flexible framework that overcomes this limitation by enabling aggressive non-orthogonal signal transmissions. In contrast to conventional non-orthogonal multiple access (NOMA) schemes, NL-COMM-Sat supports more than two UEs per receive antenna on the same frequency resource. The framework revisits optimal receiver design principles and proposes computationally efficient processing schemes that translate previously unexplored theoretical gains into tangible throughput improvements, even under realistic channel estimation errors and high-mobility Doppler conditions. Our evaluation shows that NL-COMM-Sat achieves up to a 2× increase in spectral efficiency compared to orthogonal multiple access and NOMA baselines across all considered SNR and Doppler regimes, even with a single-antenna receiver and user speeds of up to 500 km/h.
