Home // International Journal On Advances in Telecommunications, volume 15, numbers 3 and 4, 2022 // View article
Authors:
Idayat O. Sanusi
Karim M. Nasr
Keywords: Fifth Generation (5G) and beyond networks; Radio Resource Management (RRM); Distributed Algorithms; Device-to-Device Communication (D2D); Reinforcement Learning.
Abstract:
Device-to-Device (D2D) enabled cellular networks are a promising solution for Ultra-Reliable Low-Latency Communication (URLLC) systems. Integrating D2D into future wireless industrial networks and next-generation manufacturing can support the creation of massive machine-type wireless connections. In this paper, we present a Base Station Assisted (BSA) reinforcement learning approach for resource allocation in a D2D-enabled cellular network targeting smart manufacturing and Industry 4.0 applications. A distributed local Q-table is used for the D2D agents to prevent global information gathering and a stateless Q-learning approach is adopted to reduce the complexity of learning and the dimension of the Q-table. The Q-tables of the D2D agents are then uploaded to the Base Station (BS) for the resource allocation to be implemented centrally. Simulation case studies show that the presented semi distributed BSA technique results in reduced signalling overheads and a good Quality of Service (QoS) across the network compared to other conventional schemes.
Pages: 60 to 69
Copyright: Copyright (c) to authors, 2022. Used with permission.
Publication date: December 31, 2022
Published in: journal
ISSN: 1942-2601