WOS3 – Wireless, Optical and Satellite Networks
Thursday, 4 June 2026, 11:00-12:30, room Sala 5 (1st floor)
Session Chair: Lena Wosinska (Chalmers University of Technology, SE)
DL Routing Performance in DECT-2020 NR Massive Mesh Networks
Mirza Alam, Juho Pirskanen and Juha Salokannel (Wirepas Oy, Finland)
DECT-2020 NR, a global radio communication standard developed by ETSI, is designed to support industrial Internet of Things (IoT) applications. It complies with the requirements set by the ITU IMT-2020 framework. This study presents comprehensive simulations to evaluate downlink (DL) packet-forwarding performance in dense mesh networks using the DECT-2020 NR standard. Several studies address uplink performance; consequently, this specific downlink-focused research question has received limited attention in prior academic work. The study focuses on two routing mechanisms specified in the standard: Selective Flooding and Selective Source Routing, evaluated in a 1,200-node, 10-hop, large-scale deep mesh network. The simulation results indicate that the downlink (DL) capacity for Selective Flooding-where no routing information is required-is approximately 6 packets per second. For Selective Source Routing, the downlink (DL) throughput increases with the number of deterministic routing hops, reaching close to 145 packets per second when the system uses only a single 1.728 MHz channel. Both capacities are achieved while keeping the packet outage rate below 1% to comply with the IMT-2020 requirements.
Training-Free Beam Selection for 6G: A Sensing-Aided Approach
Mohammad Nabati and Toktam Mahmoodi (King’s College London, United Kingdom (Great Britain)); Subhankar Pal and Shamik Mishra (Capgemini, India)
The integrated sensing and communication (ISAC) in 6G will enable wireless networks to gain awareness of their surrounding environment. The joint design of sensing and communication introduces multiple challenges, but could also open new opportunities. Although there is a trade-off between sensing and communication, these two functionalities can also enjoy mutual benefits. In this context, sensing-aided beam selection is an area explored in the literature. Primarily, multimodal datasets containing images, GPS, LiDAR, and radar data, along with deep learning algorithms, are used in the sensing-aided beam selection process, or sensing information is used for beam tracking in line-of-sight (LoS) scenarios. However, none of the previous works have investigated the impact of static scatterers on the beam selection process using the radio sensing signals. In this paper, we propose a training-free beam selection approach, using the sensed location of scatterers. Extensive simulations show the superior performance of the proposed method compared to the traditional LoS beam in terms of signal-to-noise ratio and bit-error-rate.
Low-Complexity, Space Splitting-Based User Selection in MU-MIMO for Massive Connectivity and AI-Native Traffic
João Paulo Sales Henriques Lima, Marcin Filo, Chathura Jayawardena and Konstantinos Nikitopoulos (University of Surrey, United Kingdom (Great Britain))
The rise of Artificial Intelligence (AI)-driven services, machine-type communications, and massive Internet of Things (IoT) deployments is reshaping wireless traffic toward dense, uplink-oriented, bursty, and latency-critical patterns. In these regimes, Multi-User Multiple-Input Multiple-Output (MU-MIMO) is essential to support massive concurrent connectivity through spatial multiplexing. However, the need for frequent, low-latency scheduling decisions exposes fundamental scalability barriers in existing user selection approaches. The inherently combinatorial nature of MU-MIMO user selection leads computational complexity to grow rapidly with both the number of candidate users and spatial layers, rendering existing near-optimal heuristic methods impractical in dense and highly dynamic scenarios. This paper introduces the Space Splitting-based User Selection (SS-US) algorithm, a complexity barrier-breaking, massively parallelizable method that departs from subset-based selection by constructing orthonormal spatial bases and independently matching users to spatial directions. Simulation results across diverse MIMO configurations, channel conditions, and user densities show that SS-US reduces computational complexity by over three orders of magnitude while achieving spectral efficiency comparable to state-of-the-art practical baselines.
Energy-Efficient Learning-Based Beamforming for UAV-Satellite Communications
Arkadii Kazanskii and Suleima Briseno Ramirez (University of Luxembourg, Luxembourg); Flor Ortiz (Luxembourg Institute of Science and Technology, Luxembourg); Eva Lagunas (University of Luxembourg, Luxembourg)
Adaptive beamforming is a key enabler for UAV-satellite communications, allowing the tracking of moving satellites and the mitigation of interference in an increasingly crowded spectrum, where energy efficiency and real-time operation are critical constraints. While learning-based beamforming approaches have shown promising performance, their deployment on power-limited airborne platforms remains challenging. Motivated by the limitations of traditional beamforming methods under strict power and latency constraints, this paper proposes a codebook-based design for adaptive beamforming and evaluates its neuromorphic implementations in a representative UAV-satellite communication scenario within future non-terrestrial networks (NTNs), using neural classifiers for codebook selection. The implementation is compared with both conventional beamforming techniques and a non-neuromorphic convolutional neural network (CNN) design in terms of the SINR performance, inference latency, and energy consumption. Results reveal clear trade-offs between communication performance and hardware efficiency, showing that neuromorphic inference can significantly reduce energy consumption and latency at the cost of a slight SINR degradation.
Dynamic Switching Strategy for Cooperative Satellite-Terrestrial Networks
Nurullah Aksu and Semiha Tedik Basaran (Istanbul Technical University, Turkey); Gunes Karabulut Kurt (Ecole Polytechnique de Montreal, Canada)
The integration of Low Earth Orbit satellite constellations into 6G networks introduces significant challenges in mobility management and backhaul reliability due to frequent handovers. This paper proposes a novel cooperative dynamic switching and Group Head (GH) selection strategy utilizing a Space-Time Graph (STG) framework to enhance the reliability of Satellite-Terrestrial Networks. This framework ensures reliable backhaul connectivity by leveraging the time-varying network topology. The proposed multi-objective strategy significantly outperforms baseline methods in terms of minimizing unnecessary GH switching and optimizing operational costs. To enable this reliable performance, unlike traditional snapshot-based methods, the proposed approach employs a Heuristic Window Selection algorithm on the STG structure to perform look-ahead optimization, balancing conflicting objectives such as ground-to-satellite and inter-satellite link connectivity, and switching frequency. Additionally, an adaptive power ramping mechanism is integrated to ensure Quality of Service continuity under poor channel conditions. The study shows that transitioning the constellation into sequential formations increases the number of candidate satellites, which improves system robustness by enabling the selection of superior communication links.
