LoRaWAN-SIM

May 13, 2026 · View on GitHub

A LoRaWAN simulator for confirmed/unconfirmed transmissions and multiple gateways

List of papers where the simulator (or a variant) has been used:

D. Zorbas, S. Tulebayeva, "AI-Assistance for LoRaWAN Simulators". The corresponding paper can be found here: https://www.researchgate.net/publication/404804983_AI-Assistance_for_LoRaWAN_Simulators
D. Zorbas, C. Caillouet, K. Abdelfadeel, D. Pesch, "Optimal Data Collection Time in LoRa Networks: a Time-Slotted Approach", Sensors, Vol. 21, no. 4, 2021
D. Zorbas, "Improving LoRaWAN Downlink Performance in the EU868 Spectrum", Computer Communications, Vol. 195, Nov. 2022, pp. 303-314
D. Zorbas, "Downlink Spreading Factor Selection in LoRaWAN", Computer Communications, Vol. 215, Feb. 2024, pp. 112-119
D. Zorbas and A. Sabyrbek, "Supporting Critical Downlink Traffic in LoRaWAN", Computer Communications, Vol. 228, 107981, Dec. 2024
D. Zorbas, "LoRaWAN Network Coexistence in the EU868 Spectrum", 9th IEEE Conference on Standards for Communications and Networking, Belgrade, Serbia, Nov. 2024
S. Javed, D. Zorbas, "A LoRaWAN Adaptive Retransmission Mechanism", IEEE Conference on Standards for Communications and Networking, Munich, Germany, Nov. 2023
S. Javed, D. Zorbas, "Downlink Traffic Demand-Based Gateway Activation in LoRaWAN", 28th IEEE Symposium on Computers and Communications (ISCC), Tunis, Tunisia, July 2023
S. Javed, D. Zorbas, "LoRaWAN Downlink Policies for Improved Fairness", IEEE Conference on Standards for Communications and Networking (CSCN), Thessaloniki, Greece, Nov. 2022

Features:

EU868 or US915 frequency plans
Multiple half-duplex gateways
1% radio duty cycle for uplink transmissions
1 or 10% radio duty cycle for downlink transmissions
Two receive windows (RX1, RX2) for ACKs and commands
Non-orthogonal SF transmissions
Capture effect
Path-loss signal attenuation model
Multiple channels
Collision handling for both uplink+downlink transmissions
Proper header overhead
Node energy consumption calculation (uplink+downlink)
ADR (Tx power adjustment)
Downlink policies
Adjustable packet size and rate
AI-assisted recommendation system for improving PRR, PDR, or energy consumption

Dependencies:

Debian: apt install libmath-random-perl libgd-svg-perl libstatistics-basic-perl

Python dependencies for the recommendation system:

joblib
pandas
xgboost
tabulate

Install them with:

pip install joblib pandas xgboost tabulate

Usage:

perl generate_terrain.pl terrain_side_size_(m) num_of_nodes num_of_gateways > terrain.txt
perl LoRaWAN.pl packets_per_hour simulation_time_(hours) terrain.txt

Example with 3000x3000m terrain size, 1000 nodes, 5 gateways, 1pkt/5min, 10h sim time:

perl generate_terrain.pl 3000 1000 5 > terrain.txt
(or perl generate_terrain-m.pl 3000 1000 > terrain.txt to automatically select the number of required gateways)
perl LoRaWAN.pl 12 10 terrain.txt

JSON-based simulator input

The simulator can also be launched with a JSON configuration file. This mode is useful when the simulator is called by external tools, such as the recommendation system.

Example config.json:

{
  "packets_per_hour": 12,
  "simulation_time": 5,
  "nodes": 2000,
  "gateways": 2,
  "terrain_side": 3000,
  "number_of_bands": 2,
  "rx2sf": 12,
  "with_ack": 0,
  "max_retr": 1,
  "pkt_size": 16,
  "adr": 1,
  "double_gws": 0
}

Run the JSON-capable simulator with:

perl LoRaWAN.pl --json config.json

If the JSON file does not include a terrain file, the simulator generates one internally using generate_terrain.pl and the values of terrain_side, nodes, and gateways. The original positional-argument workflow remains supported.

AI-assisted recommendation system

The repository includes a Python launcher that runs the LoRaWAN simulator once for a baseline configuration and then uses trained ML models to recommend parameter changes. The recommendation targets are:

prr: improve Packet Reception Ratio.
pdr: improve Packet Delivery Ratio.
energy: reduce node energy consumption.

The corresponding paper can be found here: https://www.researchgate.net/publication/404804983_AI-Assistance_for_LoRaWAN_Simulators

Expected files:

LoRaWAN.pl
launcher.py
generate_terrain.pl
model/xgb_classifier.pkl
nodel/xgb_energy_regressor.pkl

Run the recommendation system with a configuration file:

python3 launcher.py --config-file config.json --target prr
python3 launcher.py --config-file config.json --target pdr
python3 launcher.py --config-file config.json --target energy

The launcher uses LoRaWAN.pl as its default Perl simulator. It writes the normalized configuration to a temporary JSON file, calls the Perl simulator with --json, parses the baseline PRR, PDR, and energy values, and then reports a ranked list of recommendations.

Useful options:

python3 launcher.py --config-file config.json --target prr --top-n 10
python3 launcher.py --config-file config.json --target energy --max-params 2
python3 launcher.py --config-file config.json --target pdr --show-baseline-output
python3 launcher.py --config '{"packets_per_hour":12,"simulation_time":5,"nodes":2000,"gateways":1}' --target prr

Notes:

Missing configuration fields fall back to the launcher's default values.
The PRR/PDR targets require the classifier model model/xgb_classifier.pkl.
The energy target requires the energy regressor model/xgb_energy_regressor.pkl.
For confirmed traffic, set with_ack to 1; for unconfirmed traffic, set it to 0.
max_retr is only meaningful when confirmed traffic is enabled.

Output sample of the basic simulator:

Simulation time = 35999.408 secs
Avg node consumption = 50.50573 J
Min node consumption = 32.32120 J
Max node consumption = 157.91968 J
Total number of transmissions = 119862
Total number of unique transmissions = 119658
Stdv of unique transmissions = 0.47
Total packets delivered = 96832
Total packets acknowledged = 0
Total confirmed packets dropped = 0
Total unconfirmed packets dropped = 22826
Packet Delivery Ratio = 0.80924
Packet Reception Ratio = 0.80924
Uplink fairness = 0.088
Script execution time = 6.2148 secs
-----
# of nodes with SF7: 197, Avg retransmissions: 0.00
# of nodes with SF8: 119, Avg retransmissions: 0.00
# of nodes with SF9: 229, Avg retransmissions: 0.00
# of nodes with SF10: 279, Avg retransmissions: 0.00
# of nodes with SF11: 159, Avg retransmissions: 0.00
# of nodes with SF12: 17, Avg retransmissions: 0.00
Avg SF = 9.135
Avg packet size = 35.912 bytes