Path Loss Prediction Accuracy Based On Random Forest Algorithm in Palembang City Area

Abstract

Path loss is a mechanism where the signal from the transmitting antenna to the receiver in a wireless network is attenuated during transmission across a medium due to external field conditions. In the telecommunication design, precise and efficient calculations are required. Random forest, as a machine learning-based path loss prediction model, is used in this study. Machine learning-based path loss prediction, random forest, has a low level of complexity and a high level of predictability. The data was collected using the drive test method at the Trans Musi busway area on the 4G network in Palembang, South Sumatra, Indonesia. The data ratio comprised 20% of the testing set and the rest of the training set. As a result, it was obtained that the prediction accuracy of 9.24% of mean absolute percentage error (MAPE) and root mean square error (RMSE) was 13.6 decibels (dB). Using hyperparameter tuning for random forest results in optimizing the model used, resulting in accuracy prediction for 8.00% of MAPE and RMSE was 11.8 dB, which is better than the previous results.