Correlation Analysis of LoRa RSSI and Climate Variables for Short-Term Forecasting Models

This study investigates the correlation between LoRaWAN RSSI (Received Signal Strength Indicator) and various climate variables by utilizing Pearson correlation, PCA (Principal Component Analysis), and XGBoost feature importance. The dataset, collected from four LoRaWAN devices and a gateway in Medellín, Colombia, encompasses diverse climate conditions over several measurements. Our analysis reveals significant positive correlations between temperature and RSSI for certain devices, while relative humidity and barometric pressure exhibit varied correlations across different nodes. The PCA visually confirms these relationships, illustrating positive correlations within the same quadrant and negative correlations in opposite quadrants. XGBoost feature importance highlights PM2.5 concentration as a crucial predictor for RSSI in one device, aligning with our correlation findings. These findings enhance the understanding of climatic impacts on RSSI, facilitating the development of more precise short-term forecasting models for LoRaWAN deployments.