Project publication · IEEE Open Journal of the Computer Society
Summary
The article applies machine learning to gateway placement in LoRa networks to reduce the carbon footprint of the deployment.
Main contribution
The proposal combines prediction and optimization models to estimate the environmental impact of different network configurations. It enables infrastructure placement decisions that consider coverage, performance and emissions.
Resultados y alcance
Los modelos de aprendizaje aproximan las soluciones de optimización exacta con un tiempo de cálculo muy inferior y conservan gran parte de la reducción de huella de carbono. El rendimiento depende de la representación espacial y de que los escenarios de entrenamiento cubran las densidades y condiciones del despliegue objetivo.
Link with BoND1
This publication reinforces BoND1 objective of designing non-cellular IoT with efficiency and sustainability criteria. The use of machine learning makes it possible to explore complex configurations without relying only on exhaustive measurement campaigns.
Project publication IEEE Open Journal of the Computer Society
Summary
The article applies machine learning to the placement of gateways in LoRa networks to reduce the carbon footprint of deployment.
Bibliometric indicators
- JCR area
- Computer Science, Theory & Methods
- JCR year
- 2025
- Scholar query
- 26 October 2026
Main contribution
The proposal combines prediction and optimization models to estimate the environmental impact of different network configurations. This allows you to decide where to place infrastructure considering coverage, performance and emissions.
Results and scope
Learning models approximate exact optimization solutions with much less calculation time and retain much of the carbon footprint reduction. Performance depends on spatial representation and whether training scenarios cover target deployment densities and conditions.
Link with BoND1
This publication reinforces BoND1’s goal of designing non-cellular IoT with efficiency and sustainability criteria. Using machine learning allows you to explore complex configurations without relying solely on extensive measurement campaigns.