Path loss prediction and evaluation at 1.8 GHz and 3.5 GHz frequencies in a densely populated environment

Abstract

Path loss (PL) models are essential for quantifying signal attenuation and ensuring reliable design and optimization of wireless communication systems. Their predictive accuracy depends on factors such as frequency, distance, and environment, making empirical validation crucial for realistic 5G deployment. This study investigates the performance of four widely used PL models; Free Space Path Loss (FSPL), Log-Distance (LD), Close-In (CI), and Floating Intercept (FI), based on extensive path loss data collection at 1.8 GHz and 3.5 GHz in Zaria, Nigeria. Results show that the FSPL model consistently underestimated path loss, yielding very high RMSE values of 44.39 dB and 42.81 dB at 1.8 GHz and 3.5 GHz, respectively. The LD model performed moderately, with RMSE values of 8.29 dB and 6.69 dB, but still outside the acceptable range for practical network planning. In contrast, both the CI and FI models produced highly accurate predictions, achieving RMSE values of 5.26 dB and 5.07 dB at 1.8 GHz, and 3.82 dB and 3.81 dB at 3.5 GHz, respectively. These findings confirm that CI and FI models are the most suitable for path loss prediction in urban environments at the considered frequencies. Findings of this work will support more efficient 5G network planning and deployment at lower frequency bands, especially in the Northern Nigeria.