Prediction of Path Loss Propagation in Case of Light Rail Transit (LRT) using Machine Learning Approaches

Path loss propagation is a very important role in the development of wireless communication networks. Path loss prediction is used for planning and optimizing the communication network. The accuracy of the path loss prediction determines the quality of the signal received by the customer. Path loss has different characteristics in different areas and different conditions. The different conditions of the transceiver, receiver, and object on the surrounding condition cause different path loss characteristics. This research aims to propose a path loss prediction model using machine learning, namely Deep Neural Network (DNN) and Support Vector Regression (SVR) in the Light Rail Transit (LRT) transportation scenario. The model is compared with the empirical path loss model, namely Okumura-Hatta and Ericsson. The dataset used is a measurement dataset in the field using the Drive Test method. This research shows that modeling using machine learning, namely DNN, has the best level of accuracy with an RMSE value of 4.155, followed by SVR, Ericsson, and Okumura-Hatta models with a value of 7.695, 40.971, and 92.079. Path loss prediction using the DNN algorithm can be used as a reference in wireless network planning.