Towards 6G: A co-simulation framework for vehicular networks in hybrid Terrestrial/Non-Terrestrial (TN-NTN) scenarios.
Fork of simu-scs (Toyota InfoTech) extended with direct Vehicle-to-Satellite (V2S) communication, dual-interface vehicle modules, and intelligent Network Selection strategies for Vertical Handover (VHO) orchestration.
The original simu-scs framework provides a Bent-Pipe Single-RAT simulation environment in which vehicles communicate via LEO satellites over a cellular backhaul. This fork extends it into a Bent-Pipe Multi-RAT architecture, enabling simultaneous 5G NR and direct satellite connectivity on each vehicle node.
| simu-scs | simu-scs-hybrid | |
|---|---|---|
| Architecture | Bent-Pipe Single-RAT | Bent-Pipe Multi-RAT |
| Vehicle Interface | Single (4G cellular) | Dual (4G/5G NR + satellite) |
| V2S Communication | ✗ | ✓ Direct Vehicle-to-Satellite |
| Network Selection | ✗ | ✓ HybridInterfaceManager + 4 strategies |
Based on the research described in:
Jing Ma, Lei Zhong, and Ryokichi Onishi, "LEO Satellite Communication Simulation Framework for Connected Vehicles," IEEE GLOBECOM 2023.
V2S Communication
V2N2V Hazard Alert Dissemination
Compound vehicle node (available in LTE variant HybridCarV2X4G and 5G NR variant HybridCarV2X5G) implementing dual-RAT connectivity:
- Cellular Interface: 5G NR protocol stack via Simu5G (
NRUe), with direct association to a gNodeB. - Satellite Interface: Direct V2S uplink/downlink in Ku-band FDD (UL 14.25/14.5 GHz, DL 10.825/11.075 GHz) via
wlan0to LEO satellites modelled by leosatellites/os3. Fleet-level channel access uses static two-channel FDMA, partitioning the vehicle fleet into two groups of 50 nodes on separate frequencies to mitigate co-channel interference. - Battery Module: State-of-Charge (SoC) tracking for energy-aware decision making.
- HybridInterfaceManager: Orchestrator module implementing the Strategy Pattern for VHO decisions. It evaluates network conditions and delegates the switching decision to one of the four pluggable strategies.
Four strategies implement the ISwitchingStrategy interface:
| Strategy | Type | Decision Criterion |
|---|---|---|
| Time-Based | Deterministic | Periodic interface alternation at a fixed interval (30 s or 60 s). |
| Coverage-Based | Reactive (Satellite-First) | Switches to satellite when geometric visibility (elevation angle) is available. |
| QoS-Based | Multi-Criteria | Weighted normalised score computed over PDR, RTT, and Jitter. Hysteresis via consecutive-degradation counting prevents oscillation. |
| Energy-Aware | Multi-Criteria | Extends QoS-Based by incorporating a battery State-of-Charge term into the utility score, balancing service quality against energy consumption. |
The original SatelliteNetworkConfigurator (leosatellites) was designed for static topologies. This fork provides an extended SatelliteNetworkConfiguratorScs that supports dynamic vehicle entry/exit via an IP pool allocator (two separate subnets: 10.3.0.0/24 for satellite, 10.1.0.0/24 for cellular) and reinstalls routing tables on topology changes.
The PositionConverter module was rewritten to auto-parse the SUMO .net.xml boundary data, enabling accurate bidirectional conversion between SUMO Cartesian coordinates and geodetic coordinates (latitude/longitude) required by the satellite propagation model.
A custom AlertServerApp (extending UdpBasicApp) implements a Vehicle-to-Network-to-Vehicle hazard alert dissemination scenario in a satellite-only environment (no terrestrial infrastructure):
- An incident vehicle (
node[0]) transmitsAlertTriggerPacketmessages to a remote server via the satellite uplink. - The server preserves the original
CreationTimeTagfrom the trigger and rebroadcasts oneAlertPacketper fleet vehicle, applying a configurable inter-send stagger (default: 1 ms) to prevent MAC queue bursts on the satellite downlink. - End-to-end latency is measured as
rcvdPkLifetime = simTime_received − triggerCreationTime, capturing the full V2N2V path.
Evaluation metrics: Uplink PDR, Fleet PDR (FPDR), Time-to-Alert (TTA), Alert Jitter.
Urban Mobility : SUMO 1.18 (TraCI co-simulation via Veins)
Simulation Core : OMNeT++ 6.0.3
Network Layer : INET 4.4.1
SUMO Bridge : Veins 5.2
Cellular Stack : Simu5G 1.2.1 (5G NR / LTE, 3GPP Rel-16)
Orbital Mechanics: OS3 (master) + leosatellites v2.0.0 (SGP4/TLE)
NTN Transport : simu-scs (Toyota InfoTech)
- OS: Linux (Ubuntu 20.04+ recommended) or WSL2
- OMNeT++ 6.0.3
- SUMO 1.18
- Git with submodule support
Framework Dependencies (auto-installed by prepare_dependencies.sh):
| Dependency | Version |
|---|---|
| INET Framework | 4.4.1 |
| Veins | 5.2 |
| Simu5G | 1.2.1 |
| leosatellites | master + v2.0.0 physicallayer |
| os3 | master |
sudo apt install -y make diffutils pkg-config ccache clang lld gdb lldb \
bison flex perl sed gawk python3 python3-pip python3-venv python3-dev \
libxml2-dev zlib1g-dev doxygen graphviz xdg-utils libdw-dev cmake wget \
openjdk-17-jre openjdk-17-jdk \
qtbase5-dev qtbase5-dev-tools libqt5svg5 qtwayland5 \
libwebkit2gtk-4.1-0 libqt5opengl5-dev \
libopenscenegraph-dev
sudo apt cleanVerify Java:
java -version # Expected: openjdk version "17.x.x"sudo nano /etc/sysctl.d/10-ptrace.conf
# Set: kernel.yama.ptrace_scope = 0
sudo sysctl --systempython3 -m pip install numpy scipy pandas matplotlib posix_ipc --break-system-packages# Download from https://github.com/omnetpp/omnetpp/releases/tag/omnetpp-6.0.3
tar xvfz omnetpp-6.0.3-linux-x86_64.tgz
cd omnetpp-6.0.3
source setenv
./configure
make -j$(nproc)
# Verify
cd samples/aloha && ./alohaNote: Run
source setenvin every new terminal, or persist it:echo "source ~/omnetpp-6.0.3/setenv" >> ~/.bashrc
git clone --recursive https://github.com/squidslab/simu-scs-hybrid.git
cd simu-scs-hybrid./prepare_dependencies.shDownloads and patches INET 4.4.1, Veins 5.2, Simu5G 1.2.1, leosatellites, and os3.
- Launch OMNeT++ IDE and create a new workspace.
- When prompted to install the INET Framework:
⚠️ DO NOT INSTALL — uncheck both options. The project requires INET 4.4.1 installed byprepare_dependencies.sh; the bundled IDE version will break the build. - File → Import → Existing Projects into Workspace
- Select
simu-scs-hybrid/as root, check all projects, click Finish. - Wait for indexing to complete before building.
Project → Build All
All scenarios reproduce a dense urban vehicular environment mapped to the Minato ward, Tokyo metropolitan area (approximately 1080 m × 1080 m, imported from OpenStreetMap via SUMO). Orbital trajectories are computed using the SGP4 propagation model from real Starlink TLE data (starlink2023.txt), with 10 satellites distributed across 2 orbital planes.
| Parameter | Value |
|---|---|
| Vehicle nodes (HybridCarV2X UE) | 100 |
| 5G Base Station (gNB) | 1 (Minato ward) |
| LEO Constellation | 10 Starlink satellites — 2 orbital planes |
| Ground Station / MCC | 1 (Fukuoka, 33.591° N 130.401° E, ≈ 900 km from Minato) |
| Evaluation duration | 400 s |
Each vehicle runs two applications:
- BackgroundLoadApp — sensor data upload to server (1024 B/packet,
exponential(500 ms)inter-packet interval) - ProbeApp — RTT measurement probe to
UdpEchoAppon server (100 B/packet,exponential(1000 ms)inter-packet interval)
The exponential inter-packet distribution models Poisson-arrival traffic, preventing artificial synchronisation artifacts that would arise with deterministic periodic transmission.
Two traffic conditions are evaluated:
| Scenario | Mean Interval | Characterisation |
|---|---|---|
| Nominal | 500 ms | Light load |
| Congested | 100 ms | Heavy load, cellular saturation |
Satellite-only scenario simulating a hazard event in an area without terrestrial infrastructure:
| Parameter | Value |
|---|---|
| Vehicle nodes | 25 |
| LEO Constellation | 10 Starlink satellites |
| Trigger send interval | 100 ms (node[0] → server) |
| Alert inter-send stagger | 1 ms (server → fleet) |
| Safety TTA threshold | 500 ms |
All commands run from:
samples-scs-hybrid/SatelliteVehicleHybridSample/HybridSCSV5GFusion/| Config | Interface | Description |
|---|---|---|
CellularOnly |
5G NR only | Terrestrial Single-RAT baseline |
SatelliteOnly |
LEO satellite only | NTN Single-RAT baseline |
SatelliteOnlyAlert |
LEO satellite only | V2N2V safety-critical scenario |
| Config | Strategy | VHO Trigger |
|---|---|---|
TimeBasedSwitching |
Time-Based | Periodic timer (30 s / 60 s) |
CoverageBasedSwitching |
Coverage-Based | Satellite geometric visibility |
QoSBasedSwitching |
QoS-Based | PDR / RTT / Jitter weighted score |
EnergyAwareSwitching |
Energy-Aware | QoS score + battery SoC |
Step 1 — Start SUMO:
sumo --remote-port 9999 --num-clients 1 -c sumodir/config.sumocfgStep 2 — Run OMNeT++ in CMDENV (recommended):
../../samples-scs-hybrid_dbg -u Cmdenv -m -c CellularOnly omnetpp.iniStep 3 — Or use the helper script:
./run.sh -c CellularOnly
./run.sh -help # list all optionsFor complete library path flags and batch run commands see
commands.sh.
Results are written to results/:
| Extension | Content |
|---|---|
.sca |
Scalar results (end-of-run statistics) |
.vec |
Vector results (time-series data) |
.vci |
Vector index |
.elog |
Event log (if enabled) |
Post-processing Python scripts (using omnetpp.scave) are provided for PDR, RTT, Jitter, QoS score, energy consumption, and V2N2V metrics.
1. Transparent (bent-pipe) payload architecture. The framework models a transparent satellite payload: the satellite relays signals between the vehicle (Service Link, Ku-band) and the Ground Station (Feeder Link, Ka-band 28.5 GHz) without on-board processing. Regenerative payload architectures where the satellite hosts an on-board gNB and the 3GPP Uu interface terminates in orbit are not supported.
2. Single-cell terrestrial segment.
The scenario uses a deliberately simplified single-cell architecture with one gNodeB. All 100 vehicle UEs are statically associated to macCellId = 1. Inter-cell Horizontal Handover is not evaluated; PDR figures for the cellular segment represent a best-case upper bound for a real multi-cell deployment.
3. Static FDMA uplink access (no deterministic scheduling). The satellite uplink divides the vehicle fleet into two static FDMA groups of 50 nodes each (Channel 0: 14.250 GHz UL; Channel 1: 14.500 GHz UL). This is a deliberate methodological choice. The resulting collision rate on the shared uplink is higher than what a TDMA/OFDMA system with centralised scheduling would produce; satellite PDR values (~90%) should therefore be interpreted as a conservative lower bound of what a real NTN access layer would achieve with deterministic channel access.
4. Break-Before-Make Vertical Handover (Hard VHO). All interface transitions, including intra-constellation Horizontal Satellite Handovers and vertical TN↔NTN transitions, follow a Break-Before-Make discipline. The vehicle disconnects from the outgoing interface before establishing the incoming one, generating a blind window during which routing table reconvergence produces packet loss. Make-Before-Break (soft handover) is a direction for future work.
This repository is part of active academic research. Bug reports and pull requests are welcome via GitHub Issues.
⚠️ This fork may contain bugs and is not an official release. It is intended for research purposes only. If you find any issues, please report them via GitHub Issues.
Inherits LGPL-3.0-or-later from the original simu-scs framework.
Copyright (C) 2023 TOYOTA MOTOR CORPORATION
SPDX-License-Identifier: LGPL-3.0-or-later
- Original Framework: simu-scs — Toyota InfoTech
- Fork Maintainer: Giuseppe Balzano — SQuIDS Lab, Università degli Studi di Napoli Federico II
simu-scs-hybrid is built as a layered composition of five independent frameworks. SUMO generates vehicle trajectories that Veins feeds into OMNeT++ via TraCI; INET provides the foundational network stack on top of which Simu5G implements the 5G NR cellular segment and leosatellites/os3 implement the LEO satellite segment.
| Framework | Version | Integration Point |
|---|---|---|
| INET | 4.4.1 | Foundation layer: IP stack, physical layer models, node mobility, UDP/TCP |
| os3 | master | SGP4 orbital mechanics and NORAD TLE parsing for satellite trajectory computation |
| leosatellites | v2.0.0 (phy) + master | LEO constellation physical layer and SatelliteNetworkConfigurator (extended in this fork) |
| Veins | 5.2 | Bidirectional SUMO–OMNeT++ co-simulation bridge via TraCI protocol |
| Simu5G | 1.2.1 | 5G NR / LTE protocol stack — NRUe is the base type for HybridCarV2X5G |