Work Packages 3

Overview

Work Package 3 (WP3) centers around the development of Giulia, an open-source, Python-based, system-level simulator for Integrated Non-Terrestrial Networks (iNTNs). Giulia is designed to model both ground UEs (GUEs) and aerial users like UAVs, supporting 4G, 5G, and emerging 6G networks across terrestrial, aerial, and spaceborne layers. Its goal is to provide an advanced simulation tool for research and optimization of next-generation networks.

1. Simulator Engine and Architecture

Event-Driven Simulation Engine (Task 3.1)

  • Giulia uses an event-driven simulation approach, where events are scheduled, executed, and rescheduled with a custom calendar.

  • Key advancements:

    • Support for object-oriented events instead of just variable passing.

    • Dynamic rescheduling and compatibility with external AI algorithms like reinforcement learning.

    • Major code efficiency improvements for faster simulations.

    • Rigorously tested in diverse multi-layer network scenarios.

  • Task Status: Now considered complete.

System-Level Architecture & GUE Playground (Task 3.2)

  • A modular architecture was introduced, with configurable blocks (e.g., antenna setup, UE mobility, resource allocation, performance monitoring).

  • Includes over 25 predefined templates for common deployment scenarios (urban macro, micro, NTN).

  • Now supports PyTorch tensor operations, enabling GPU acceleration and faster processing of complex scenarios.

  • Supports:

    • Multi-layer simulation (4G/5G/6G).

    • cm-wave band (10–17 GHz).

    • Hotspot & aerial highway modeling.

    • Priority-based cell reselection algorithms.

    • Dynamic energy consumption models with sleep modes and carrier shutdowns.

  • The simulation process follows three phases: Initialization, Dynamic Configuration, and Execution & KPI Collection.

2. UAV and NTN Integration

UAV Integration (Task 3.3)

  • Integrated 3GPP-compliant modules for urban UAV operations (based on TR 36.777).

  • Models include:

    • UAV-specific mobility, altitude, antenna configs.

    • Channel effects: LoS, path loss, fading, shadowing.

  • Introduced aerial highway modeling, reflecting future UAV traffic frameworks.

  • Future work will further refine UAV models using real-world measurement data.

Aerial and Spaceborne Integration (Task 3.4)

  • Integrated LEO satellite modules with accurate deployment and channel models (rain/cloud/gas/scintillation).

  • Newly added support for HAPS (quasi-stationary platforms at ~20 km), including atmospheric and wind effects.

  • Enables full simulation of hybrid networks: terrestrial + aerial + spaceborne.

3. Energy and Performance Optimization

  • Giulia includes advanced power consumption modeling for all BS types (4G–6G), with:

    • Static/dynamic power components.

    • Real-time updates based on traffic demand.

    • Energy-saving strategies like beam deactivation and carrier shutdown.

4. Calibration and Validation

  • Giulia has been calibrated against 3GPP and ITU standards, showing strong alignment in:

    • SINR, RSRP, and coupling loss.

  • Includes a regression testing framework to validate updates and ensure consistency.

  • Calibration examples provided for 6 industry scenarios (e.g., UMa, UMi, 3GPP TR benchmarks).

5. ML-Based Channel Modeling (Task 3.5)

  • Not yet started, but groundwork is in place.

  • Measurement campaigns in Benidorm and Valencia gathered RSRP data.

  • Future work will train machine learning models for high-fidelity aerial channel simulations, incorporating:

    • Spatial/temporal variations.

    • Altitude and interference dynamics.

Conclusion and Next Steps

Giulia has evolved into a state-of-the-art, open-source simulator for iNTNs, offering:

  • Multi-layer, multi-technology simulation.

  • Energy modeling and dynamic resource optimization.

  • Integrated support for UAVs, LEO, and HAPS.

  • GPU acceleration and ML compatibility.

Next steps include:

  • Expanding 6G features.

  • Enhancing traffic and resource algorithms.

  • Continuing GPU customization.

  • Advancing ML-based aerial channel models.

Volver arriba