Soft Clustering for Enhancing ITU Rain Model based on Machine Learning Techniques

Abstract

With the many folds increase in demand for capacity in mobile broadband communication technology every year, wireless carriers must be prepared for the tremendous increase in mobile traffic in coming years. It forces scientists and researchers to come up with new wireless spectrum bands which has capabilities to support higher data rates. The higher spectrum bands like millimeter waves are the candidate band for this type of problems. This band comes with the challenges of radio wave attenuations oof signals due to the presence of gases, water vapor and other weather phenomenon like rain, storms, snow, hail etc. Different models are presented in order to predict attenuation due to rain out of which ITU-R model is the widely acceptable model. The ITU-R model contains complex methodology for calculating regression coefficients which are depends on frequency and polarization. In this paper, K-Means algorithm is used to propose an improved ITU-R model. Proposed model can make up the shortcoming of ITU-R model to determine the break-up points in frequency range and obtained soft clusters have been trained by machine learning algorithms then proposes a mathematical model for prediction of radio wave attenuation due to rain. The implementation results of proposed model were also compared with the ITU-R model.