A Novel Reinforcement Learning Based Adaptive Optimization of LTE-TDD Configurations for LTE-U/WiFi Coexistence

Abstract

In order to meet the exponential rise in mobile data demand, it is proposed to supplement existing licensed bandwidth with the unlicensed spectrum, where a part of cellular data demand is served in the unlicensed spectrum. This deployment of LTE technology in unlicensed spectrum is termed as LTE in unlicensed spectrum (LTE-U). However, deploying LTE technology in the unlicensed 2.4/5 GHz band will greatly cause interference to the already existing technologies such as WiFi and Zigbee due to the stark differences in their channel access mechanisms. For LTE-U to become a reality, it is necessary for LTE to fairly coexist with the existing technologies. Therefore, in this paper, we propose a Q-Learning based Dynamic Frame Selection (DFS) algorithm which ensures fair coexistence between LTE-U and WiFi technologies. Furthermore, we propose the use of reduced power subframe to limit interference to the co-channel users and increase the channel utilization. By extensive simulation we show the effectiveness of our proposed DFS algorithm when compared to existing algorithms in the literature.