Machine Learning-Based DRX Mechanism in NR-Unlicensed

Abstract

Continuous increase in number of wireless devices and data rate demand has motivated the usage of 5G New Radio in Unlicensed band (NR-U). Coexistence between 5G NR-U with other unlicensed technologies like Wi-Fi is challenging owing to the reduced probability of channel availability. Waiting to get access to unlicensed channel results in more battery consumption of User Equipment (UE). Discontinuous Reception (DRX) mechanism in NR-U can reduce UE’s energy consumption. In this letter, we introduce Machine Learning (ML) based DRX mechanisms. The DRX is modeled through three state semi-Markov model. We utilized the ML classifiers to get access to the unlicensed channel. The simulation results point out that the performance of LSTM is superior to the Support Vector Machine (SVM) and Naive Bayes (NB) classifiers, which achieves up to 3% improvement in Power-Saving Factor ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">PSF</i> ) and 50% reduction of average delay compared to conventional 5G-DRX.