Study on radio propagation prediction by machine learning using urban structure maps

In recent years, mobile data traffic has been increasing, and high-quality mobile communication services are required. Therefore, it is essential to understand the complex radio propagation characteristics in an actual communication environment. In this paper, we propose a method of predicting radio propagation characteristics by machine learning called gradient boosting, in which the feature is the building information around a transmission point and a receive point, which greatly affects radio propagation characteristics. Gradient boosting is a method of constructing a prediction model using a plurality of weak learners, and it is possible to output the importance of the input feature. That is, it is possible to quantify which feature had a large impact on the radio propagation prediction, and to verify the validity of the model. We evaluated the prediction accuracy using the measured data in an urban area, and clarified the effect of the difference of the feature on the prediction accuracy.