Q-Learning Based Fair and Efficient Coexistence of LTE in Unlicensed Band

Abstract

The increased demand for spectrum resources for multimedia communications and a limited licensed spectrum have led to widespread concern regarding the operation of long term evolution (LTE) in the unlicensed (LTE-U) band for internet of things (IoT) systems. Because Wi-Fi and LTE are diverse with dissimilar physical and link layer configurations, several solutions to achieve an efficient and fair coexistence have been proposed. Most of the proposed solutions facilitate a fair coexistence through a discontinuous transmission using a duty cycling or contention mechanism and an efficient coexistence through a clean channel selection. However, they are constrained only by fairness or efficient coexistence but not both. Herein, we propose joint adaptive duty cycling (ADC) and dynamic channel switch (DCS) mechanisms. The ADC mechanism supports a fair channel access opportunity by muting certain numbers of subframes for Wi-Fi users whereas the DCS mechanism offers more access opportunities for LTE-U and Wi-Fi users by preventing LTE-U users from occupying a crowded channel for a longer time. To support these mechanisms in a dynamic environment, LTE-U for IoT applications is enhanced using Q-learning techniques for an automatic selection of the appropriate combination of muting period and channel. Simulation results show the fair and efficient coexistence achieved from using the proposed mechanism.