Forecasting Throughput Capacity on 5Ghz Wireless Radio Network Using Linear Regression Method

Abstract

In an internet network, internet service providers must ensure the quality of the devices used is in good condition. One of the devices used is a 5GHz wireless radio where the stability of the fluctuating throughput capacity affects the quality of the device as a network transmission. This research uses machine learning with a simple linear regression model to predict the final throughput capacity based on the initial throughput and optimize it in the future. The regression model is used because it has a good accuracy level in forecasting. The data used is time series throughput data from the Ubnt Litebeam 5AC Gen.2 5GHz wireless radio device for the period of January-July 2024, using data split technique, data from January-June is used as training data and data from July as test data. The forecasting results obtained indicate the model's error rate measured using Mean Square Error (MSE) of 184,682 and Root Mean Squared Error (RMSE) of 13,589. From the forecasting results obtained, it can be used as a reference to evaluate the feasibility of wireless radio devices used by internet service provider for the future.