Analysis of Millimeter Wave Path Loss Prediction using Machine Learning Techniques

In this paper, different machine learning algorithms like ridge regression, linear regression, Random forest regression and K-Nearest Neighbors Algorithm (KNN) were used to predict the path loss of millimeter waves (mmWave)under different scenarios like Uma (Urban Macro) and Umi (Urban Micro) thereby comparing the accuracy of each algorithm. mmWaves are prone to attenuation due to different environmental factors like foliage, size and rate of raindrops, etc. as the size of these objects are comparable to the wavelength of the mmWaves. Since mmWave is the basis for 5G communication it is imperative to analyze and predict the path loss exponent under different scenarios. From the analysis performed it is seen that linear regression provides better accuracy compared to the other models.