Scientific Objective (SO) 1: iNTN coexistence and optimization
As of December 2023, our research contributed approximately 35% towards the goal of creating a new theoretical—but practical—framework for designing and optimizing large-scale iNTNs for UAV connectivity.
As a summary, our research so far in this front, detailed in the paper «Designing Cellular Networks for UAV Corridors via Bayesian Optimization,” has developed an innovative methodology for cellular network design, addressing the needs of both GUEs and UAVs flying in aerial corridors. Leveraging Bayesian Optimization (BO), a sophisticated machine learning technique, we maximized UAV SINR, thus enhancing their reliability. Our methodology rooted in both theoretical frameworks and practical applications, seeked to provide insights into the re-engineering of existing cellular infrastructures for robust 3D connectivity. Highlighting the advantages of designated aerial highways, which facilitate regulated UAV flights, our research has proposed an advanced framework for the optimization of electrical antenna tilts and transmit power at BSs. This is pivotal in significantly improving UAV performance within these corridors. The key achievements of our work include:
Demonstrating a substantial 23.4 dB gain in mean SINR for UAV corridors compared to a baseline configuration, underscoring the practical viability of our framework in bolstering UAV connectivity.
Effectively preserving GUE performance with a slight gain in SINR, thus ensuring that optimization for UAV corridors does not adversely affect ground-based cellular services.
Introducing a scalable optimization framework capable of adapting to various performance functions, thereby offering the flexibility needed in network design to embrace future advancements and requirements.
Our research marks a significant step forward in tackling the intricate challenge of integrating UAVs into cellular networks, utilizing BO to navigate the trade-offs between ground and aerial connectivity effectively.
Looking ahead, we aim to refine and expand our optimization framework to further explore its applications across various scenarios. Additionally, we plan to showcase the transfer learning capabilities of our BO optimizers. Our goal is to continue advancing the development of sophisticated network infrastructures that can meet the evolving demands of UAV connectivity and support the ambitious objectives of future communication networks, including 6G.
Fundamental performance limits and trade-offs of an iNTN
In this front, we achieved approximately 40% of our scientific objective, which aims to unveil the fundamental performance limits and trade-offs of an integrated iNTN in terms of coverage, reliability, latency, capacity, and energy efficiency across various network layers and system configurations.
Our work so far, summarized in the recent study «Throughput and Coverage Trade-Off in Integrated Terrestrial and Non-Terrestrial Networks: An Optimization Framework,» represents a significant milestone in our journey. This research dived into the complex dynamics between coverage and throughput within iNTNs, proposing an innovative mathematical framework to balance and understand the trade-offs among these critical performance metrics effectively. In more details, our framework leveraged advanced mathematical modelling and optimization to identify optimal configurations that maximize network coverage and/or GUE throughput across terrestrial and non-terrestrial layers. Our analysis sheds light on the inherent performance limits of iNTNs, highlighting how distinct network configurations can impact key performance indicators. By systematically exploring these configurations, we provided a comprehensive understanding of the potential and limitations of iNTNs in supporting diverse communication needs. Our comprehensive analysis exposes the suboptimality of current network configurations, revealing that our optimized approach can surge mean data rates by over 200% compared to networks lacking satellite integration and those conforming to standard 3GPP system model recommendations. GUE outages can also be greatly reduced at expense of affordable data rates. This stark improvement underscores the framework’s capacity to transform coverage and capacity, especially crucial for expanding reliable connectivity in remote regions.
Our findings offer valuable insights for policymakers and network operators, suggesting strategies for deploying and managing iNTNs to achieve desired performance outcomes. Moving forward, we plan to further refine our optimization framework and explore its application in different types of scenarios and expand to other metrics. By continuing to investigate the performance limits and trade-offs of iNTNs, our goal is to contribute to the development of robust, efficient, and scalable network infrastructures that can support the ambitious objectives of 6G.
Significant strides have been made toward the fulfilment of Scientific Objective 2, achieving approximately 27% of progress towards our outlined goals. This progress is notably anchored in two pivotal research contributions that address the challenges of energy efficiency and advanced mobility within the emerging paradigms of 5G and 6G telecommunications.
As a summary, and with respect to energy efficiency, our innovative research, detailed in the publication «High Altitude Platform Stations: The New Network Energy Efficiency Enabler in the 6G Era,» marks a significant advancement in the quest for energy-efficient radio access networks (RANs). This study brings to light the indispensable role of non-terrestrial networks, particularly HAPS, in mitigating the rising energy demands faced by conventional terrestrial RANs amid escalating wireless traffic. We introduced a comprehensive quantitative framework designed to meticulously assess and compare the energy consumption profiles of HAPS-integrated RANs against traditional terrestrial network setups, utilizing real-world traffic patterns as a benchmark. The centrepiece of our strategy is the development of an innovative traffic offloading algorithm, which facilitates the strategic redirection of terrestrial network traffic to HAPS. This mechanism allows for the temporary deactivation of ground-based coverage macrocells during low traffic periods, thereby reducing unnecessary energy consumption. The deployment of HAPS, characterized by their self-sustainability, elevated operational altitudes, and extensive coverage areas, together with our algorithm, has demonstrated potential energy savings of nearly 30%. These findings not only elevate our understanding of energy efficiency in the 6G era, but also contribute significantly to the sustainable evolution of wireless networks. Our future main endeavours will extend this research to include satellite nodes, further broadening the scope of our energy-saving solutions.
Parallel to our efforts in energy efficiency, our exploration of advanced mobility, interference, and beam management has led to the development of a groundbreaking approach, as elaborated in «A Novel Metric for mMIMO Base Station Association for Aerial Highway Systems.» This research emphasizes the critical need for enhancing drone communications within 5G and 6G networks, focusing specifically on UAVs navigating through aerial highways. By introducing a novel metric tailored for mMIMO BSs, we address the complex challenges of aerial mobility, including channel correlation and dynamic interference, thereby facilitating more efficient and reliable UAV communication links. This metric, distinguished as the Eigen score, transcends the limitations of traditional RSRP methods by incorporating advanced considerations such as spatial multiplexing and signal quality, ensuring UAVs are connected to the most suitable BS. Our research highlights the inadequacies of RSRP in providing optimal connectivity for UAVs, particularly in high-altitude conditions where channel correlation and line of sight interference significantly impact communication efficacy. The Eigen score metric has proven to enhance network efficiency, reliability, and drone operational capabilities within aerial highway systems, with evaluations showing improvements in UAV signal quality by up to 3.30dB. As we advance, our research will delve deeper into refining this metric and exploring its application to mMIMO beam optimization across large networks, aiming to bolster the autonomy and effectiveness of UAVs in the interconnected landscape of future wireless networks.
Scientific Objective 3: Open source iNTN system level simulator
We have reached a pivotal milestone in our project, achieving approximately 30% of our objective. This success is marked by the completion of the first edition of Giulia, a cutting-edge system-level simulator that stands as a testament to our commitment to advancing UAV-able iNTN network research. The simulator’s capabilities are as follows:
Accurate Interaction Modelling: Giulia adeptly models the interactions between terrestrial and non-terrestrial nodes, offering invaluable insights into the complex dynamics serving both terrestrial and non-terrestrial users, such as drones. This feature is critical for understanding the nuanced communication patterns and optimizing network performance in mixed-use environments.
Advanced Cellular Feature Support: The simulator supports essential cellular technologies, including multicarrier operation and mMIMO with a grid of beams. These features are vital for assessing the network’s ability to handle high data volumes and provide reliable connectivity across diverse scenarios.
Giulia is engineered on an event-driven framework, enabling it to simulate the intricate network dynamics accurately. It captures the evolution of traffic and communication channels over time, integrating realistic traffic and channel models. By abstracting the physical (PHY) layers and simulating the medium access control (MAC) layers of 4G/5G technologies, Giulia provides a comprehensive view of network behavior, including coverage, capacity, and energy consumption.
Developed in Python, Giulia offers an extensive range of templates for established simulation scenarios, adhering to standards such as ITU.R.M2135, 3GPP 36.814, 3GPP 38.901, and 3GPP 36.777. This versatility ensures that Giulia can accommodate various research needs, facilitating the exploration of new theories and the validation of innovative network algorithms.
Giulia has been uploaded to GitHub and has been registered as a result of the Universitat Politècnica de València. Our progress towards the complete development and testing of the proposed UAV-aware network algorithms and the overall iNTN optimization framework is on track, with significant milestones already achieved in simulator functionality and scenario modelling.
Summary of Work Package 2: UAV-Aware Network Algorithm Design
Overview
Work Package 2 focuses on the design of advanced algorithms to support UAV-aware and NTN-integrated 5G/6G networks. It addresses load redistribution, mobility management, interference coordination, and beam steering, with a strong emphasis on enabling reliable aerial connectivity for Unmanned Aerial Vehicles (UAVs) while maintaining the performance of ground users (UEs). The work integrates terrestrial networks, Low Earth Orbit (LEO) satellites, and High Altitude Platform Stations (HAPS).
1. Load Redistribution and Multi-Connectivity (Task 2.1)
Although limited progress was made recently, prior simulations showed that offloading traffic to HAPS can yield major energy savings by enabling terrestrial BSs to enter sleep mode. These results, published in «High Altitude Platform Stations: the New Network Energy Efficiency Enabler in the 6G Era», lay the foundation for future enhancements.
Next Steps:
Incorporate LEO satellite BSs into the framework to optimize energy-aware traffic offloading.
2. Mobility Management (Task 2.2)
This task focuses on optimizing handover (HO) processes for UEs, initially on the ground and later extended to UAVs. The framework targets the optimization of:
Time-to-Trigger (TTT): Delay before initiating a HO.
A3 Offset: Signal strength threshold for switching cells.
Bayesian Optimization—particularly the TuRBO algorithm—is used to balance conflicting KPIs:
Handover Ping-Pongs (HPP)
Handover Failures (HOF)
Radio Link Failures (RLF)
Key Findings:
Optimizations effectively reduced HOFs and HPPs in single-speed scenarios.
In mixed-speed scenarios, higher weighting on HPP improved overall results.
Optimizing for RLF instead of HOF led to better SINR tail performance and 25% fewer ping-pongs.
Next Steps:
Simulate more realistic environments with multiple streets and speeds.
Apply transfer learning to adapt HO models across speed ranges and extend to UAVs at 50 m and 150 m altitudes.
3. Coordinated Interference Management (Task 2.3)
UAVs at high altitudes suffer from increased interference due to LoS with multiple BSs and reception from antenna sidelobes. A data-driven framework was developed to optimize:
Antenna tilt and
Vertical Half-Power Beamwidth (vHPBW)
Model Setup:
19 sites with 3-sector BSs in a 500 m hexagonal grid (Urban Macro model).
UAVs fly at 150 m along aerial highways; ground UEs remain at 1.5 m.
TuRBO (Trust Region BO) outperformed other methods (e.g., SAASBO, VSBO) by:
Improving UAV SINR by 21 dB.
Reducing outages from 100% to nearly 0%.
Increasing UAV data rates by 71% (from 730 kbps to 1.25 Mbps).
Minimizing impact on ground UEs (only 3% throughput loss).
Next Steps:
Use city-scale maps and ray tracing for realistic simulations.
Apply transfer learning to adapt antenna configs across different aerial heights.
4. Efficient Beam Steering (Task 2.4)
To combat the challenges of LoS interference and poor MIMO performance in UAVs, the team developed a massive MIMO beam steering framework that intelligently configures Synchronization Signal Block (SSB) beams.
Optimization Workflow:
Aerial Highway Segmentation (via PSO):
Splits UAV flight path into segments and assigns optimal serving cells using a new metric: MAMA (mMIMO-Aerial-Metric-Association).
MAMA factors include signal strength, spatial diversity, and interference.
Beam and Power Optimization (via elite Genetic Algorithm):
Selects SSB beams and allocates power to maximize SINR.
Includes constraints to preserve ground UE performance.
Results:
SINR Gains: 5%-tile SINR for UAVs improved by 5.15 dB.
UAV Data Rate Gains:
5%-tile data rate increased from 2 Mbps to 11 Mbps.
Mean rate improved from 12 Mbps to 32 Mbps (166%).
Ground UE Impact: Only a 3.1% drop in ground UE throughput, demonstrating strong aerial-terrestrial balance.
Next Steps:
Extend beam optimization techniques to NTNs:
Enable dynamic beam steering from moving LEO satellites and HAPS.
Evaluate performance through NTN-specific simulations for connectivity and interference management.
Conclusion
WP2 achieved significant advancements in UAV-aware network design. Across its four core tasks, it developed robust optimization frameworks to manage:
Traffic and energy efficiency via HAPS/LEO integration.
Seamless handovers with adaptive parameter tuning.
Interference reduction through smart antenna configuration.
Precision beam steering leveraging massive MIMO.
The techniques—especially TuRBO and MAMA—show strong potential for scalable deployment in complex 6G environments integrating UAVs, satellites, and terrestrial networks.
