Title: Assign Hysteresis Parameter For Ericsson BTS Power Saving Algorithm Using Unsupervised Learning

Abstract: Gaza Strip suffers from a chronic electricity deficit that affects all industries including the telecommunication field, so there is a need to optimize and reduce power consumption of the telecommunication equipment. In this paper we propose a new model that helps GSM radio frequency engineers to choose the optimal value of hysteresis parameter for Ericsson BTS power saving algorithm which aims to switch OFF unused frequency channels, our model is based on unsupervised machine learning clustering K-means algorithm. By using our model with BTS power saving algorithm we reduce number of active TRX by 20.9%.