5G-Optimizing Network Coverage in Radio Self Organizing Networks by M/L Based Beam Tilt Algorithm

Abstract

This paper proposes a novel machine learning based antenna beam tilt algorithm for minimizing the overall Poor Signal Strength (PSS) regions /dead zones in the network area considered. Our objective is to provide network intelligence and automation of the optimization of the configurable parameter, azimuth angle of the antenna, to adapt to varying channel conditions and rebalance the entire network so as to provide an optimized level of service to the users. The proposed scheme involves developing a simulation scenario for the existing network and employing machine learning to study the behavior of the network by taking large number of combinations of azimuth angles and corresponding measure of PSS area. Regression analysis and stochastic gradient descent are used to obtain the relationship and the optimized angles for which the PSS area is minimum. Our simulation results demonstrate the reduction in overall PSS area compared to state of art approaches.