Empirical Evaluation and Fine-Tuning of Path Loss Models for LoRaWAN Propagation

This study quantifies the accuracy of the theoretical models to determine how closely each model predicts the measured path loss and identify the most suitable model for LoRaWAN deployment in Edo State environment, Nigeria. LoRaWAN, a wireless communication technology based on LoRa modulation, relies on accurate path loss predictions to estimate network coverage. Path loss estimation is typically done using empirical propagation models, but these models are often based on urban environments with specific characteristics. Given that the accuracy of these models can vary significantly depending on environmental factors, it is essential to assess their performance in different settings. A quantitative experiment was conducted, involving the measurement of LoRaWAN propagation characteristics at various locations around a base station situated at an elevation of 8m. The measured path loss data were obtained across representative urban, suburban and open space environments, and was compared with predictions from the Free Space Path Loss, Okumura–Hata, and Log-Distance models. In the study significant underestimations of path loss by all models were identified, with a discrepancy of approximately 32.8dB for the Log-Distance Model. To improve model accuracy, the research suggests fine-turning and the introduction of a correction factor of 13.93dB and path loss exponent of 3.62, reducing the root mean square error to 29.8dB. These findings provide valuable insights for optimizing LoRaWAN deployment in similar environments and highlight the need for further adaptation of empirical models for more accurate path loss prediction in diverse regions.