AI4C5 – AI/ML Solutions for Communications
Thursday, 4 June 2026, 16:30-18:00, room Sala 2 (1st floor)
Session Chair: Carlos J. Bernardos (Universidad Carlos III de Madrid, ES)
Meta-Learning-Based Service-Aware User Associations: An ORIGAMI Approach
Michail Kalntis (Delft University of Technology, The Netherlands); Andres Garcia-Saavedra (NEC Laboratories Europe, Germany); Andra Lutu (Telefónica Research, Spain); George Iosifidis (Delft University of Technology, The Netherlands)
Future 6G networks are expected to support highly heterogeneous and latency-critical services over dynamic radio and compute infrastructures spanning the edge-to-cloud continuum. In this setting, user association can no longer be treated as a purely radio-centric problem, as end-to-end performance depends jointly on radio access conditions and service-specific latency, both of which may vary rapidly over time. Based on the architecture of the EU-funded project ORIGAMI and its Dynamic Fairness and Load Balancing (FLB) use case, this paper studies a service-aware user association problem under non-stationary conditions, where a controller assigns each user to a serving cell while keeping the application-dependent (from transportation, processing, etc) and potential handover (if there is a change in association decisions) delays to a minimum. Importantly, these latencies are not only unknown at the time the user associations are determined but are also assumed to be time-varying and even chosen by an adversary aiming to disrupt the user’s quality of experience/service. To solve this problem and address these challenges, we propose a framework based on online meta-learning that combines the decisions of multiple learners (or algorithms, experts), each with different learning rates (as the problem is difficult to tackle) via a meta-learning algorithm that tracks and chooses the best performing one(s) for the experienced conditions. Finally, we establish a theoretical dynamic regret guarantee against a powerful benchmark (i.e., oracle) and validate the approach using real-world data from a top-tier mobile network operator in Europe.
Semantic MIMO: Revisiting Linear Precoding in the Generative AI Era
Chunmei Xu, Yi Ma and Rahim Tafazolli (University of Surrey, United Kingdom (Great Britain))
This paper revisits linear precoding, namely match-filter (MF) and zero-forcing (ZF), in a semantic multiple-input multiple-output (MIMO) system empowered by generative AI. The aim is to examine whether interference, channel state information (CSI) accuracy, and scalability limitations in conventional MIMO systems remain critical. Theoretical analysis based on the generative inference model and Lipschitz continuous assumptions reveals reduced sensitivity to interference and channel imperfections, as well as performance inferiority in high-SINR regimes compared to conventional MIMO systems. Simulation results validate the analysis and show that MF achieves semantic performance comparable to ZF under both perfect and imperfect CSI. These findings suggest that semantic MIMO relaxes the needs for aggressive interference mitigation and highly accurate CSI, while improving scalability with reduced computational and implementation complexity.
Topology-Aware Deep Reinforcement Learning for Multi-Hop URLLC in DECT-2020 NR Mesh Networks
Benjamin Rauwolf, Awais Bin Asif and Jürgen Peissig (Leibniz Universität Hannover, Germany)
DECT-2020 New Radio (NR) is a radio interface technology (RIT) standardized by the European Telecommunications Standards Institute (ETSI) to support massive machine-type communication (mMTC) and Ultra-Reliable and Low-Latency Communications (URLLC) in large-scale mesh deployments. DECT-2020 NR supports multi-hop communication, typically by flooding the packet, which increases network load and does not explicitly account for constraints such as end-to-end reliability and end-to-end latency. In this paper, we propose a URLLC-based multi-hop communication framework that utilizes a topology-aware Deep Reinforcement Learning (DRL) policy for DECT-2020 NR mesh networks. The key idea is to model next-hop selection as a Markov decision process that uses link- level latency and packet error rate (PER) metrics consistent with the DECT-2020 NR system model. An inductive graph neural network (GNN) encoder summarizes the mesh topology and a per-edge DQN head, trained in a doubling fashion, selects next hops using a URLLC-shaped reward that couples end- to-end latency and end-to-end reliability. Simulation results for small to large mesh topologies show that the proposed solution, i.e., GNN+DRL multi-hop communication policy, improves the URLLC trade-off compared to the existing approach and DECT- 2020 NR standard flooding procedure: it reduces end-to-end latency up to 19.9% and end-to-end PER up to 10% for a medium-sized mesh network at a given reliability target and reduces the number of transmissions per delivered packet in comparison with the existing approach.
LLM-ASF6G: LLM-Based Algorithm Selection Framework for 6G Network Service Optimization
Anestis Dalgkitsis, Cyril Shih-Huan Hsu, Chrysa Papagianni and Paola Grosso (University of Amsterdam, The Netherlands)
The integration of cloud computing into mobile networks has unlocked new possibilities through a unified and abstract infrastructure. However, this convergence introduced significant complexity, emphasizing the critical need for innovative service management and orchestration techniques. Despite considerable efforts to incorporate automation for optimizing network services, no single algorithm can achieve optimal performance across all spatial and temporal conditions. To address this challenge, we propose LLM-ASF6G, a novel algorithm selection framework for 6G network service orchestration, designed for algorithmic optimization across both pre-deployment planning and post-deployment operational phases. The proposed framework dynamically selects the most suitable optimization algorithm based on real-time performance metrics and user-defined preferences from a diverse pool of algorithms, techniques and strategies. We evaluate the performance of LLM-ASF6G using different LLM to assess its feasibility for real-world deployment and implementation.
Design and Evaluation of a Multi-Agent Perception System for Autonomous Flying Networks
Diogo Ferreira (Faculdade de Engenharia da 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)
Autonomous Flying Networks (FNs) are emerging as a key enabler of on-demand connectivity in dynamic and infrastructure-limited environments. However, current approaches mainly focus on UAV placement, routing, and resource management, neglecting the autonomous perception of users and their service demands-a critical capability for zero-touch network operation.
This paper presents the Multi-Agent Perception System (MAPS), a modular and scalable system that leverages multi-modal large language models (MM-LLMs) and agentic Artificial Intelligence (AI) to interpret visual and audio data collected by UAVs and generate Service Level Specifications (SLSs) describing user count, spatial distribution, and traffic demand. MAPS is evaluated using a synthetic multimodal emergency dataset, achieving user detection accuracies above 70% and SLS generation under 130 seconds in 90% of cases. Results demonstrate that combining audio and visual modalities enhances user detection and show that MAPS provides the perception layer required for autonomous, zero-touch FNs.
