On the Accuracy and Efficiency of Received Signal Strength Modelling for a Forest Environment

Within telecommunications, the accuracy and efficiency of machine learning models (ML) define their utility, and consequently, the choice of ML mechanisms assumes paramount importance. This study focuses on the exploration and comparison of diverse ML and ensemble learning techniques, with a specific emphasis on their significance in crafting precise and extensive models. To this end, the quality of the received signal and the optimization and functioning of wireless communication networks rely heavily on accurately predicting the received signal strength indicator (RSSI) and path loss (PL). The studied environment, which is highly complex, spans 2000 km2 of the intricate landscapes of the American River Hydrologic Observatory (ARHO) networks and is characterized by a diverse blend of terrain features and vegetation distributions. Notable independent variables under consideration include path distance, canopy coverage, terrain variability, and path angle. The proposed ensemble ML approaches demonstrate remarkable accuracy and efficiency when it comes to modeling and predicting the RSSI values in forested environments.