Contextual Bandits Approach for Selecting the Best Channel in Industry 4.0 Network

Abstract

Ultra dense heterogeneous network of new radio in unlicensed band (NR-U) is a key technology for potentially accomplishing the capacity and seamless connection goal of next-generation wireless communication systems in Industry 4.0 network. Such deployment results in the cell proliferation with diverse cell size in overlapping condition, which leverage various channel connectivity option for the NR-U user. Nevertheless, coexistence of several other NR-U and/or legacy unlicensed band users in the common channel is a major technical challenge to be resolved, which severely degrades the user’s quality of service (QoS). Hence, this paper is based on the channel selection functionality for mobile NR-U users to select the best channel to use for uplink transmission in the unlicensed band. We model this problem using contextual bandits as the set of context information is provided to the user. We use Thompson’s sampling algorithm to solve the problem. The simulation result has been presented to show the effect of noise on the performance of our proposed approach.