Machine Learning Enabled Dynamic Spectrum Access for 6G Wireless Networks

Abstract

In this article, a 6G scenario is considered which allows sharing of spectrum between links which have spectrum license (6G license access), and which are unlicensed (Wi-Fi). A machine learning algorithm is formulated which considers both, medium access and physical layer parameters of access points and eNodeBs to (i) enable them to select sub-channels, and (ii) permit them to choose best sub-channel in a distributed manner. The simulation results demonstrate that (i) proposed method converges to ideal scenario, and (ii) throughput of both, licensed and unlicensed systems remains close to results obtained through a search which has been conducted exhaustively.