Advancing the Future of Integrated 5G-Satellite Networks: A Practical

Framework for Performance Evaluation, Dataset Generation, and

AI-Driven Approaches

Najmeh Alibabaie

, Antonello Calabr

, Pietro Cassar

Alberto Gotta

and Eda Marchetti

CNR-ISTI, Via Moruzzi 1, Pisa, Italy

Keywords:

Joint Simulation, Network Simulator, LEO Satellite Communications, Satellite-Terrestrial Integrated

Networks, Ray Tracing, Back-Hauling.

Abstract:

This paper introduces a framework for Satellite, Terrestrial Integrated Network (STIN), a modular and joint

simulation tool for simulating and evaluating integrated terrestrial and non-terrestrial communication systems.

The framework comprises various modules designed to model real-world environments, compute and analyze

constellation features, and perform channel modeling. Through the seamless integration of these components,

the STIN framework enables users to assess the performance of satellite constellations under diverse conditions

and select optimal conﬁgurations for enhanced coverage and communication efﬁciency. The paper discusses

the methodology and workﬂow of the framework and a preliminary implementation, suggesting avenues for

obtaining communication datasets to support AI-driven approaches.

1 INTRODUCTION

Combining 5G and satellite networks is important for

future global communication systems. In situations

where regular networks don’t work well or are hard

to use, this combination becomes crucial. Some ex-

amples are during disasters, in remote places, or in

areas like the ocean or air where normal networks

can’t reach. User groups like rescue teams, mobile

users, and critical infrastructure operators need reli-

able communication systems in these situations.

In response to these challenges, Satellite-

Terrestrial Integrated Networks (STINs) presents a

promising solution by merging the high capacity and

low latency of terrestrial 5G with the wide coverage

and resilience of satellite networks, particularly those

utilizing Low Earth Orbit (LEO) mega-constellations.

In hybrid architectures, satellites provide remote cov-

erage and act as a backup plan to ensure continuous

telecommunication services for critical applications.

https://orcid.org/0009-0002-8281-9767

https://orcid.org/0000-0001-5502-303X

https://orcid.org/0000-0002-3704-4133

https://orcid.org/0000-0002-8134-7844

https://orcid.org/0000-0003-4223-8036

However, the integration of terrestrial and non-

terrestrial networks (NTN) poses signiﬁcant technical

challenges, such as high propagation delays in satel-

lite links, on-and-off connection due to satellite mo-

bility, the need for seamless handover management

between terrestrial and satellite systems, and interfer-

ence mitigation in shared frequency bands. Addition-

ally, protocol adaptation across different layers, from

the physical layer to the core network, is necessary to

maintain reliable end-to-end performance within the

hybrid network.

Indeed, the complexity of the multidimensional

aspects that must be considered in the real-world ex-

perimentation of STINs encourages the development

of simulation approaches and environments to design

and analyze STIN performance and quality. Accurate

simulation environments allow researchers to assess

whether the designed non-terrestrial network meets

speciﬁc application requirements across multiple lay-

ers, including the physical, network, and application

layers. However, existing simulation tools often do

not fully address the challenges posed by STINs or

lack comprehensive end-to-end analysis capabilities.

Some are also not well-suited to interact with various

existing simulation platforms.

312

Alibabaie, N., Calabrò, A., Cassarà, P., Gotta, A. and Marchetti, E.

Advancing the Future of Integrated 5G-Satellite Networks: A Practical Framework for Performance Evaluation, Dataset Generation, and AI-Driven Approaches.

DOI: 10.5220/0013564400003970

In Proceedings of the 15th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2025), pages 312-319

ISBN: 978-989-758-759-7; ISSN: 2184-2841

Recent real-world projects, including SpaceX

Starlink’s

experiments for 5G backhaul, AST Space-

Mobile’s

direct-to-device satellite service trials, and

OneWeb’s

collaborations with mobile operators for

hybrid 5G satellite connectivity, emphasize the grow-

ing demand for simulation-based performance eval-

uation tools. Furthermore, the 3GPP Release 17

guidelines for NTN integration provide a standardized

baseline that this proposed simulation framework can

utilize to ensure relevance for future system designs.

This paper aims to ﬁll this gap by presenting

a simulation-driven evaluation framework for inte-

grated 5G-satellite networks, incorporating cross-

layer modeling and multi-tool integration. By focus-

ing on key performance indicators (KPIs) such as end-

to-end latency, throughput, handover success rates,

and efﬁciency in mobility management, the proposed

framework seeks to provide both new insights into the

feasibility and optimization of future STIN deploy-

ments and useful datasets.

Therefore, our ﬁst objective is to simulate a STIN

network, focusing on the methodology and a prelim-

inary implementation of integrating tools that, while

not fully capable on their own, can be combined to

effectively simulate and analyze real-world scenarios.

Although the tools used in our initial implementation

are commercial, the methodology remains valid when

replacing each tool with an open-source alternative

that offers similar capabilities. This ﬂexibility allows

readers to adapt our approach to their needs and re-

sources.

The second objective of our methodology is to

provide datasets generated through our simulation

framework. These datasets can help researchers over-

come obstacles encountered in real-world environ-

ments, enabling the analysis and prediction of vari-

ous features through Artiﬁcial Intelligence (AI) and

Machine Learning (ML)-driven approaches. Recog-

nizing the current lack of comprehensive and high-

quality datasets in the ﬁeld, our objective is to con-

tribute to the development of a framework that not

only facilitates the generation of valuable data for fu-

ture STIN research but also promotes the sharing of

these datasets among researchers. In the long term,

this may reduce reliance on proprietary tools, ulti-

mately fostering advancements in the ﬁeld through

data-driven insights and collaborative research ef-

forts.

The remainder of this paper is organized as fol-

lows. In Section II, a brief review of the related work

on simulation tools applied in STIN research is pre-

https://satellitemap.space/

https://ast-science.com/

https://oneweb.net/

sented. In Section III, we delve into the methodology

of our proposed STIN simulation framework, detail-

ing its design, components, and capabilities that make

it well-suited for simulating STINs. In Section IV, we

provide a validation example of execution to show-

case the performance of our proposed framework in

a real-world scenario. In Section V, we conclude the

paper by summarizing our key ﬁndings and contribu-

tions and shedding light on the potential future work.

2 RELATED WORK

Nowadays, several advanced simulation tools have

been developed and utilized to evaluate the perfor-

mance of STINs. These simulators are broadly cat-

egorized based on several key factors. For instance,

some tools are speciﬁcally designed to support mega-

constellations, which consist of large groups of satel-

lites in low Earth orbit working collaboratively to en-

hance connectivity and coverage.

Additionally, these simulation tools vary in their

availability. Some are offered as commercial soft-

ware, often providing extensive support and spe-

cialized features for professional use, while others

are available for free, promoting accessibility for re-

searchers and developers. Furthermore, each simula-

tor comes with distinct applications and functionali-

ties, allowing users to adapt their analyses to speciﬁc

scenarios, whether for network optimization, perfor-

mance testing under varied conditions, or assessing

the impact of environmental factors on satellite con-

nectivity.

Without being exhaustive, this section provides an

overview of the tools commonly used to analyze satel-

lite constellations and terrestrial communication sys-

tems. More detail on existing solutions, their evalu-

ation, and comparison can be found in (Jiang et al.,

2023), (Yastrebova et al., 2021).

NS-2, NS-3: NS-2, an older discrete event simu-

lator, can simulate various network protocols across

wired and wireless networks, including satellite net-

works. However, it lacks a graphical user interface,

presenting challenges in usage. NS-3 addresses this

limitation with its user-friendly Python interfaces and

data analysis tools (Puttonen et al., 2021), (Sormunen

et al., ). As an improvement over NS-2, NS-3 offers

better modularity, extensibility, and more realistic and

accurate models for various network components.

Matlab: it offers comprehensive toolboxes, including

5G and Satellite Communications Toolboxes, making

it well-suited for physical layer and link-level simula-

tions in STINs. Its alignment with 3GPP Release 17

and ready-to-use NTN channel models ensures that

Advancing the Future of Integrated 5G-Satellite Networks: A Practical Framework for Performance Evaluation, Dataset Generation, and

AI-Driven Approaches

313

researchers can easily study waveform adaptation,

propagation effects, and initial performance evalua-

tion. The visualization capabilities and integration

with Simulink

further enhance multi-layer modeling

across physical, MAC, and network layers. However,

Matlab is a commercial tool and could be expensive

when a combination of multiple specialized toolboxes

is required. Scalability could be an issue in the case

of large-scale network-level simulations (e.g., thou-

sands of satellites and devices). Additionally, its built-

in support for higher-layer protocols (e.g., transport

and application) could require customization or ex-

ternal integration. Finally, its real-time performance

and hardware-in-the-loop capabilities are less ma-

ture compared to dedicated network simulators; there-

fore, for end-to-end, large-scale STIN performance

evaluation, a hybrid simulation approach combining

Mathlab with other event simulators is recommended

(Mannoni et al., 2022).

System Tool Kit (STK): STK is a powerful simu-

lation tool that facilitates the construction and anal-

ysis of satellite constellations, exploration of air and

spacecraft missions, and modeling of hybrid network

performance. It is particularly useful for simulating

physical layer performance metrics based on satellite

propagation models, path loss models, and antenna

and transceiver models. However, STK targets mainly

the system-level mission analysis rather than detailed

communication protocol simulation. Integration with

external tools like Matlab or ns-3 can be required for

simulating end-to-end communication stacks or data

link, transport, and application layers (Li et al., 2021).

Hypatia: Hypatia is a framework designed for simu-

lating and visualizing Low Earth Orbit (LEO) constel-

lations. It combines NS-3 and the Cesium 3D map-

ping library, enabling the simulation of satellite tra-

jectories, link utilization changes, and available band-

width changes over time (Kassing et al., 2020). A

notable limitation of Hypatia is its lack of ﬂexibility

in constructing various scenarios and its constrained

visualization capabilities.

Space Networking Kit (SNK): SNK is a network-

ing platform tailored for LEO mega-constellations.

It allows users to easily construct complex scenarios

through conﬁguration ﬁles and a single bash com-

mand, facilitating the evaluation and visualization

of communication processes (Wang et al., 2024).

While recent enhancements improve SNK’s capa-

bilities, they may increase complexity for inexperi-

enced users and demand additional computational re-

sources, potentially affecting simulation efﬁciency for

users with limited hardware.

Satellite Network Simulator 3 (SNS3): SNS3 is a

https://mathworks.com/products/simulink.html

modular and ﬂexible satellite model built upon the

open-source Network Simulator 3 (NS-3). It incor-

porates DVB-S2 and DVB-RCS2 speciﬁcations for

forward and return links, respectively, making it a

scalable and adaptable open-source simulator for net-

working research and development (Puttonen et al.,

2015).

OS3 OMNET++: OS3 is an open-source satel-

lite simulator built on OMNET++, offering modular-

ity, extensibility, and adaptability for simulating di-

verse satellite constellations and applications (Valen-

tine and Parisis, 2021). Despite the clear documenta-

tion and tutorials available for OMNET++, the lack of

speciﬁc resources for OS3 might pose challenges for

users seeking guidance on on its unique features and

capabilities. This limited support may prevent users

from fully utilizing OS3’s potential for various simu-

lation scenarios.

QualNet: QualNet is commonly used for modeling,

simulating, and analyzing the performance of com-

munication networks, particularly in scenarios where

communication endpoints are constantly changing

their position relative to each other or ﬁxed infrastruc-

ture (Jennings et al., 2010).

Gpredict: Gpredict is a Linux-based program

that provides real-time satellite tracking and orbit

prediction. It utilizes the SGP4/SDP4 propagation

algorithms and NORAD TLE to achieve this func-

tionality (Jennings et al., 2010). As a Linux-based

program, Gpredict might not be accessible to users

on other operating systems without additional setup

or virtualization.

As highlighted in this not exhaustive survey, ex-

isting simulation solutions are very heterogeneous in

the provided features and the target KPIs. Therefore,

in designing a uniﬁed platform for STINs servers, is-

sues should be considered, such as i) discrepancies

in programming languages, network structures, and

data formats across different tools; ii) using software

conceived for work either with satellite or terrestrial

networks independently; iii) integrating these diverse

networks by combining the strengths of multiple sim-

ulation platforms; iv) complements the existing solu-

tions with specialized capabilities to smoothly solve

communication and data format issues.

If some of these issues have already been suc-

cessfully targeted in speciﬁc scenarios, such as indoor

communications environments (Hussain et al., 2024),

research is still needed to deﬁne a platform that can

effectively analyze the communication and channel

characteristics in integrated satellite-terrestrial net-

works. Therefore, the target of the paper is to com-

bine the strengths of various simulation platforms to

SIMULTECH 2025 - 15th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

314

overcome the complexities of integrating diverse net-

works and contribute to more comprehensive STIN

evaluations.

3 METHODOLOGY

In this section, we introduce the methodology used

for developing our simulation framework designed for

STINs. This framework integrates different tools to

offer a modular solution for simulating various satel-

lite constellations and communication scenarios. Fig-

ure 2 depicts the STIN simulation framework. Our in-

tegrated STIN framework consists of three main com-

ponents, similar to those found in most software. The

input part involves feeding the framework with the

initial data necessary for analysis and obtaining re-

sults. This part is responsible for providing the simu-

lation with essential data and conﬁgurations, allowing

users to customize parameters and input the required

information for network representation. The core part

processes the input data, closely observing the con-

stellation to determine the distribution and movement

of satellites to create an accurate representation of the

real environment

Additionally, it generates a 3D version of the

ground that closely resembles the real environment.

Finally, it visualizes the communication between the

obtained constellation and the ground. The output

part of our framework presents the simulation events

Figure 2: The STIN simulation framework.

and results, offering users useful information to ex-

plore, understand, and assess the performance of

STIN under different scenarios and conﬁgurations.

To schematize the ﬂow of data between different

tools in our framework, we provide a sequence dia-

gram (Figure 1) of the methodology execution using

three commercial tools (Matlab, Blender, and Wire-

less InSite (WI)). As mentioned in the introduction,

the selected tools were chosen due to their availabil-

ity within our institution. Alternative tools with simi-

lar performance can also be utilized.

In this case, the diagram illustrates the step-by-

step interactions between the main components of

our proposed STIN framework, highlighting how the

modules collaborate to simulate and analyze satellite

constellations in a realistic environment. The user-

Figure 1: Sequence Diagram of proposed STIN simulation.

Advancing the Future of Integrated 5G-Satellite Networks: A Practical Framework for Performance Evaluation, Dataset Generation, and

AI-Driven Approaches

315

provided inputs are transferred to different tools based

on their requirements. All data necessary for constel-

lation analysis and data related to the simulation un-

der investigation are transferred to MATLAB, where

the core processes related to that are executed. Any

data required for generating 3D models of the ground

environment is sent to and handled in Blender, which

performs the relevant processes. The data obtained

from these tools, along with other parameters, are

sent to Wireless InSite (WI) to compute the com-

munication performance between terrestrial and non-

terrestrial components. Finally, the simulation events

and results are sent back to the framework for visual-

ization and further analysis by the user. Considering

the Core component of Figure 2, in the following sec-

tions, details of each of its modules are provided.

3.1 Data Collection Module

This module simulates a satellite communication sce-

nario over time by iterating through several steps. As

shown in Algorithm 1, it ﬁrst deﬁnes the scenario’s

details, including date/time, locations, and the satel-

lite constellation. Then, for each time step, it retrieves

the current location of the ground node and collects

a FeatureDataset containing the Date/Time, Azimuth,

Elevation, and Distance for the entire constellation.

The function repeats this process until the simulation

is complete and returns the collected FeatureDataset

for further analysis.

Algorithm 1. Collect-Data (TimeDetails).

1: Set Step ← 1

3: while Step ≤ TotalTime / SampleTime do

4: Get the current location of Ground Node

5: Get the FeatureDataset for the entire constellation

(Date/Time, Azimuth, Elevation, Distance)

6: Increment Step

7: end while

8: return FeatureDataset

3.2 Analyser Module

The Analyzer module processes the collected features

from the scenario simulation and analyzes the satellite

constellation. Based on the Algorithm 2, in line 1, it

begins by calculating the Line-of-Sight (LOS) satel-

lites based on antenna orientation, which determines

which satellites are within the line of sight of the

ground nodes. Then in line 2, it computes the Mass

distribution function for LOS satellites’ Elevation,

Azimuth, and Distance to estimate the probability dis-

tribution of these parameters.

The function then determines the count of non-

line-of-sight locations in line 3 and calculates the

number of time intervals without visibility (line 4). In

line 5, it calculates the minimum and maximum satel-

lites in the line of sight for each point on the ground,

as well as the Access Interval - statistics of visibility

duration of LOS satellites, including the minimum,

maximum, and average values in line 6.

Based on line 7, the function identiﬁes satellites

with comprehensive coverage, which helps locate

satellites that provide the most extensive coverage to

the ground nodes. Finally, it visualizes the analy-

sis to represent the ﬁndings of the satellite-ground

node communication, such as coverage, visibility, and

other essential characteristics (line 8). The function

concludes by returning the calculated satellite statis-

tics, providing valuable information for further anal-

ysis or decision-making.

Algorithm 2. Consellation-Analyzer (FeatureDataset).

1: SatStatistics (1) ← Calc-LOS ()

2: SatStatistics (2) ← Cal-Mass ()

3: SatStatistics (3) ← Comp-NLOS-count ()

4: SatStatistics (4) ← Calc-visibilityless ()

5: SatStatistics (5) ← Calc-LOS-Stat ()

6: SatStatistics (6) ← Calc-AcInterval-VisDuration ()

7: SatStatistics (7) ← Ident-Comp-Coverage ()

8: Visualize (SatStatistics)

9: Return SatStatistics

3.3 Selection and Export Module

The Selection and Export module performs satellite

selection and data export for simulation purposes. In

Algorithm 3,it takes the FeatureDataset and SatStatis-

tics, ground location as input. In line 1 it selects rep-

resentative satellites based on the analysis of the con-

stellation. In line 2, it calculates the ground projec-

tion of the selected satellites on the ground and ﬁnally,

it returns information about transmitter and receivers

for further use in simulations.

Algorithm 3. Selection-AND-Export ( Feature-Dataset,

Sat-Statistics, Ground-data).

1: [El, Az, Dis] ← Select-Representative ()

2: Satloc ← Calc-Projection (El, Az, Dis)

3: Return SatLoc

3.4 Geometry Extraction Module

As depicted in Figure 2, the Core component also in-

cludes modules managing the geometry and the chan-

nels. The Geometry Extraction module is a cru-

cial component for streamlining the CAD prepara-

tion process in simulating real-world environments

using Blender. It integrates the Blosm3 add-on, which

simpliﬁes the creation of 3D CAD models for sim-

ulation purposes. Utilizing Blosm3’s functionality,

SIMULTECH 2025 - 15th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

316

the module enables the following tasks: download-

ing real-world terrain data to ensure realistic simula-

tions; generating 3D CAD models of buildings from

the OpenStreetMap dataset; placing CAD models on

the terrain while considering various building height

and ﬂoor options; and importing forests and individ-

ual trees as 3D objects to enhance the simulation en-

vironment’s realism.

3.5 Channel modeling Module

This module focuses on modeling the interaction of

transmitted rays with surrounding geometries through

reﬂections, diffractions, and transmissions. Channel

models are incorporated to enable a realistic evalu-

ation of integrated terrestrial and non-terrestrial net-

work solutions’ performance.

Algorithm 4. Channel-Modeling ( Geometries, WaveDe-

tails, AntennaDetails, TransmittersLoc, ReceiversLoc).

1: Geo3D ← Import (Geometries (Train, City, Foliage))

2: SetMaterials (Geometries)

3: TransceiverWave ← DeﬁneWave (WaveDetails)

4: TransceiverAnten ← DeﬁneAntenna (AntennaDetails)

5: Trans ← EstablishTransmitter (TransceiverWave,

TransceiverAnten, TransmittersLoc)

6: Receiv ← EstablishReceiver (TransceiverWave,

TransceiverAnten, ReceiversLoc)

7: SetStudyArea (Geo3D, Trans, Receiv)

8: SetCommunicationSystem ()

9: RunSimulation ()

10: Return (Result)

The location of the transceivers has been imported

in line 1, and the geometric characteristics of the ge-

ometries’ material, waves, antennas, transmitters, and

receivers are deﬁned in lines 2 to 6. Ray-tracing mod-

els have been employed in this module to simulate

the scenario, facilitating a detailed analysis of the

network’s performance within the given study area,

which is determined in lines 7 and 8. After running

the simulation (line 9) with the speciﬁed settings, the

module returns the results for further analysis.

4 VALIDATION OF EXECUTION

To validate our proposed framework, we applied it

in a simulation involving a real-world satellite con-

stellation and a deﬁned terrestrial area, demonstrat-

ing its practicality and performance under realistic

conditions. Speciﬁcally, we used the TLE (Two-

Line Element) ﬁle of the Starlink constellation

and

set the simulation to begin at 18:30:00 local time on

https://celestrak.org/NORAD/elements/

Figure 3: Access analyses between the satellites and the

ground station using Matlab.

2025/02/22, with a total duration of 20 minutes and

a sampling interval of 5 seconds. The terrestrial ob-

servation grid covered a rectangular area bounded by

latitudes 43.7061 to 43.7277 and longitudes 10.3834

to 10.4357, representing a set of ground nodes for vis-

ibility analysis.

During the constellation observation phase, the

framework monitored satellite movement and ground

coverage over time. This allowed the system to record

link availability and other spatial-temporal metrics

relevant for connectivity assessment. Figures 3 illus-

trates the simulated satellite constellation and ground

area setup. Figure 5 displays a portion of the dataset

collected after executing Module 1, and were then

passed to the next module for analysis.

Key features of the constellation were evaluated

using the analysis module. Figure 6 presents a sum-

mary of the analyzed features, which contributed to

the subsequent selection of an optimal set of con-

stellations with the most coverage in the simulation

area. Based on the extracted metrics, the simulation

recorded 197 unique satellites observed during the de-

ﬁned time window. Each ground node maintained

LOS with between 11 and 25 satellites, and no re-

ceiver was left without visibility throughout the ob-

servation period.

Furthermore, the visibility duration per satellite

ranged from 5 to 150 seconds, offering insight into

temporal coverage dynamics. These results guided

the realistic selection of satellite subsets and their rel-

ative positions. With the analyzed features and other

inputs, the next component selected an optimal set

of constellations that provided the most comprehen-

sive coverage in the simulation area. Figure 7 show-

cases the selected satellites and their projected loca-

tions on the ground. To accurately reﬂect the physi-

cal environment in our simulation, we conﬁgured and

extracted the geometry of the target area bounded by

latitudes 43.7061 to 43.7277 and longitudes 10.3834

Advancing the Future of Integrated 5G-Satellite Networks: A Practical Framework for Performance Evaluation, Dataset Generation, and

AI-Driven Approaches

317

Figure 4: 3D CAD model of simulation area (Blender).

Figure 5: The output of Collect-Data module (Matlab).

Figure 6: The output of Consellation-Analyzer module

(Matlab).

to 10.4357 using Geometry Extraction Module. Fig-

ure 4 illustrates the resulting 3D representation of the

simulation environment.

After completing the constellation selection and

projection and geometry extraction, the data, consist-

ing of transmitter and receiver locations in a CSV ﬁle,

along with DAE ﬁles containing terrain, buildings, fo-

liage, and other objects, is fed into the channel mod-

eling module. Figure 8 depicts the channel modeling

process, which generated results that could be used

for further post-processing or AI-driven analysis.

For each ground node, we compute a normal-

ized ray reception ratio, deﬁned as the number of

rays received with power above the NB-IoT receiver

sensitivity threshold divided by the total number of

rays traced to that node. These results are shown in

Figure 9, where each cell corresponds to a speciﬁc

Figure 7: The output of Selection-AND-Export module

(Matlab).

Figure 8: Visualize several propagation Rays obtained in

the communication (WI).

ground location (deﬁned by latitude and longitude)

and indicates the link reliability or ray coverage qual-

ity at that point. Higher values (close to 1) indicate

strong and consistent reception, while lower values

suggest potential outages or weak coverage areas due

to obstacles or poor link geometry. The generation

of these comprehensive datasets through our frame-

work paves the way for various data-driven applica-

tions in the ﬁeld of STIN. These datasets can be used

to develop, train, and test ML and AI tasks such as

coverage prediction, outage detection, adaptive UAV

placement, or dynamic beam steering. Furthermore,

researchers use them to explore advanced applica-

tions, such as reinforcement learning for autonomous

network management and deep learning for enhanced

signal processing and interference mitigation.

5 CONCLUSION AND FUTURE

WORK

Our STIN framework presents a comprehensive and

modular approach for simulating and evaluating satel-

lite constellations in realistic environments. By in-

corporating various modules for observation, analy-

sis, and channel modeling, the framework offers valu-

SIMULTECH 2025 - 15th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

318

10.3834

10.3909

10.3983

10.4058

10.4133

10.4208

10.4282

10.4357

Longitude

43.7061

43.7115

43.7169

43.7223

43.7277

Latitude

0.6257

0.8543

0.6163

0.9914

0.9657

0.9714

0.8

0.9371

0.8576

0.72

0.8971

0.8161

0.7018

0.2743

0.4919

0.1057

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Figure 9: Normalized Link Quality per Ground Location in

the Simulation Area.

able insights into the performance and behavior of in-

tegrated terrestrial and non-terrestrial communication

systems. In future work, we will focus on further

enhancing our STIN framework. Key areas of im-

provement include reﬁning the efﬁciency of constel-

lation selection algorithms and integrating advanced

AI techniques for more in-depth data analysis. These

advancements will enable more accurate and efﬁcient

simulation and evaluation of integrated 5G-satellite

networks. Additionally, we will prioritize the devel-

opment of an automated platform to facilitate seam-

less communication and information exchange be-

tween various simulation tools. This automation will

streamline the simulation process, making it more ac-

cessible and efﬁcient for a broader range of users.

Furthermore, recognizing that reliance on commercial

tools may not be feasible for all researchers, we will

continue to explore alternative simulation tools and

platforms. Our goal is to create a framework com-

posed entirely of free and open-source tools, ensuring

that the proposed methodology can be implemented

and utilized by anyone, regardless of their access to

commercial software.

ACKNOWLEDGEMENTS

We thank Tarun Chawla and Remcom Inc for the

Wireless InSite X3D ray tracer numerical analysis.

This work is supported by RESTART (PE00000001)

under the PNRR of the Italian MUR program

NextGenerationEU.

REFERENCES

Hussain, S., Bacha, S. F., Cheema, A. A., Canberk, B.,

and Duong, T. Q. (2024). Geometrical features

based mmwave uav path loss prediction using ma-

chine learning for 5g and beyond. 5:5667–5679.

Jennings, E. H., Segui, J. S., and Woo, S. (2010). Ma-

chete: Environment for space networking evaluation.

In AIAA Int. Conf.on Space Operations, page 1–12.

Jiang, W., Zhan, Y., Xiao, X., and Sha, G. (2023). Network

simulators for satellite-terrestrial integrated networks:

A survey. 11(0):98269–98292.

Kassing, S., Bhattacherjee, D., Aguas, A. B., Saethre, J. E.,

and Singla, A. (2020). Exploring the internet from

space with hy patia. In Proc. of the ACM Int. Measure-

ment Conference, page 214–229. New York, USA.

Li, J., Hua, N., Zhao, C., Zhu, K., Li, Y., , and Zheng, X.

(2021). Design and implementation of open optical

satellite network emulation platform (oosn-ep) based

on distributed multi-node system. In Opto-Electronics

and Communications Conf., page 1–3. Hong Kong.

Mannoni, V., Berg, V., Cazalens, S., and Raveneau, P.

(2022). System level evaluation for nb-iot satellite

communications. In IEEE 95th Vehicular Technology

Conference, page 1–6. Helsinki, Finland.

Puttonen, J., Herman, B., Rantanen, S., Laakso, F., and

Kurjenniemi, J. (2015). Satellite network simulator

3. workshop on simulation for euro pean space pro-

grammes. In Workshop on Simulation for Eu Space

Programmes, page 26. Noordwijk, Netherlands.

Puttonen, J., Sormunen, L., Martikainen, H., Rantanen, S.,

and Kurjenniemi, J. (2021). A system simulator for

5g non-terrestrial network evaluations. In IEEE 22nd

Int. Symposium on a World of Wireless, Mobile and

Multimedia Networks, pages 292–297.

Sormunen, L., Huikko, T., R

onty, V., Sepp

anen, E.,

Rantanen, S., Laakso, F., and Puttonen, J. Simu-

lative comparison of dvb-s2x/rcs2 and 3gpp 5g nr

ntn technologies in a geostationary satellite scenario.

In 12th Adv. Satellite Multimedia Syst. Conf., page

arXiv:2502.13704.

Valentine, A. and Parisis, G. (2021). Developing and exper-

imenting with leo satellite constellations in omnet++.

Wang, X., Han, X., Yang, M., Han, S., and Li, W. (2024).

Space networking kit: A novel simulation platform for

emerging leo mega-constellations. In IEEE Int. Conf.

on Communications, pages 5590–5595. Denver, USA.

Yastrebova, A., Anttonen, A., Lasanen, M., M., Vehkaper

, and H

oyhty

a, M. (2021). Interoperable simulation

tools for satellite networks. In IEEE 22nd Interna-

tional Symposium on a World of Wireless, Mobile and

Multimedia Networks, pages 304–309.

Advancing the Future of Integrated 5G-Satellite Networks: A Practical Framework for Performance Evaluation, Dataset Generation, and

AI-Driven Approaches

319