Sionna-SAGIN

This repository contains a simulation and evaluation framework for a UAV-assisted Space-Air-Ground Integrated Network (SAGIN) architecture designed to improve Direct-to-Cell (D2C) communication performance in diverse terrains. The code leverages realistic propagation and fading models to simulate signal quality, throughput, latency, and error rates for baseline and UAV-relayed architectures.

Introduction

Traditional D2C communication suffers from high propagation losses, capacity bottlenecks, and limited reliability in remote or disaster-stricken areas. This framework evaluates a SAGIN-based solution where UAVs serve as adaptive relays between LEO satellites and user terminals (UTs), decoupling ionospheric and multipath effects. We consider three communication topologies:

• Trad-D2C: direct satellite-to-UT links
• SAGIN-D2C: relayed via UAVs at fixed positions
• Opt-SAGIN-D2C: relayed via UAVs with optimized 3D placement

Realistic terrain-aware shadowing and fading models are used across Desert, Mountain, and City environments.

Setup and Requirements

• Python 3.8+
• Jupyter Notebook
• Required packages:
  tensorflow, sionna, optuna, numpy, scipy, matplotlib, seaborn, pandas

Usage

Launch Jupyter and open the evaluation_notebook.ipynb.

Execute all cells to simulate performance over various terrains and topologies.

The notebook outputs:

CSV file with SNR, BLER, latency, and throughput results

Plots for comparative analysis in results/ and plots/ directories

Citation

This work was submitted to MILCOM 2025:

Abdullah Al Noman, Talip Tolga Sarı, Sunday Amatare, Gokhan Secinti, and Debashri Roy, "Space-Air-Ground Network for Direct-to-Cell Communication," IEEE Military Communications Conference (MILCOM), October 2025 [Accepted].