Ray Tracing-Based LoRaWAN Gateway Placement for Reliable Connectivity in Amazonian Regions

Network planning is an important task in wireless communications, as it helps network operators avoid unnecessary costs. In the context of the internet of things, using long-range wide-area network technologies in the Amazon rainforest, a key challenge is ensuring reliable communication between end-devices and gateways (GWs). In this sense, this reliability is strongly affected by channel conditions. Thus, during the planning phase, choosing the appropriate channel model is an important decision for accurate simulations. Given this motivation, in this work, we propose an optimization model to evaluate the impact of different types of channels on coverage and packet delivery ratio in a forest scenario. We used channels from ray tracing, empirical, and stochastic approaches to assess how decisions made during the network planning phase, in terms of the channel used, affect GW placement and, specifically, the percentage of end-devices covered and the reliability of the communication system. Our results show that GW placement based on site-independent channels can overestimate the number of gateways required to meet the network requirements, whereas using site-specific channels allows us to satisfy the same requirements with fewer GWs.

3D scenario

The 3D scenario files used in this work can be accessed by this link. In this folder, there is files that can be manipulated using Blender software.

Compiling ns-3 LoRaWAN

The first step to use our framework, is compiling the ns-3 LoRaWAN module. You can follow the steps defined in this link to do this.

✍️ Installing the Python environment for optimization

Afte compile the ns-3 module, you need to configure python environment for Sionna RT. To do so, you need to execute the following command to install the conda environment

conda env create -f environemnt.yml

Generating stochastic or emprical channel using ns3-LoRaWAN

To generate stochastic or empirical channel models, you should move the source code lorawan_general_energy_simulation.cc to the ns-3-dev/scratch. After that in the ns-3-dev/scratch folder, you should run the following command:

./ns3 run scratch/lorawan_general_energy_simulation.cc -- --spreadingFactor=7 --channelType=cost --scenario=forest

The available flags are:

• --spreadingFactor: The number of spreading factor, vary between [7, 12].
• --channelType: Type of channel to used. The options are log-distance, Okumura-Hata, COST-231, Nakagami, two ray, 3gpp-UMa, WI (x3D), WI (Full 3D), Sionna. To use one of these channel you should use the following options: log, okumura, cost, nakagami twoRay, threegpp, rural, wix, wif, sionna.
• --simulationTime: Time of the simulation.
• --scenario: Type of grid organization used. The availabe options are: etoile, forest, and canyon.

:gear Gateway placement optimization

In order to perform a gateway placement optimization, with a fixed threshold you can use the following command:

python3 gateway_positioning -c okumura -t -120

In this case, an optimization process will running with a scenario with Okumura-Hata channel and a power threshold of -120 dBm. Furthermore, the following flags can use to change the behavior of the optimization:

--channel: Type of channel can be okumura, cost, rural, log, wif, wix.

--threshold: Power threshold in dBm.

If you would like to perform an optimization using an interval of power threshold, you can use the following command:

python3 multi_rho_gateway_positions -c cost --max-rho -80 --min-rho -150

In this case, the power threshold interval consider a minimum power of -150 dBm and maximum power of -80, with a COST-231 channel. Furthermore, the following flags can use to change the behavior of the optimization:

--channel: Type of channel can be okumura, cost, rural, log, wif, wix.

--threshold: Power threshold in dBm.

--max-rho: Maximum threshold power.

--min-rho: Minimum threshold power.

📊 Result plots

plot_position.py: Plot the grid with each position obtained from the optimization model.

plot_pdr.py: Plot packet delivery ratio (PDR) bar char for different channel models.

plot_time.py: Plot bar chart related to the simulation time to obtain physical-level metrics for different channel models.

plot_multi_threshold.py: Plot line chart related to the number of gateways suggested for different received power thresholds.

plot_ed_grid.py: Plot the end-devices grid organization across different scenarios.

plot_tradeoff.py: Script to generate a scatter plot for depicting the relationship between the MSE of received power (considering WI received power as ground truth) and the simulation time.

plot_receiver_power.py: Script to plot the CDF of received power of all end-devices across different channel models.