Adaptive Network Management in 6G-Assisted Industrial IoT via Multi-Objective Deep Reinforcement Learning
Abstract
The fast development of the Industrial Internet of Things (IIoT) system during the era of 6G communications requires intelligent and adaptable network management systems to address the highly dynamic and resource-constrained environment. Conventional network optimization methods are not able to pursue several conflicting goals at the same time e.g. latency, throughput, energy efficiency and reliability. In order to address this deficiency, this paper puts forward a Multi-Objective Deep Reinforcement Learning (MODRL)-based framework of adaptive network management of 6G-assisted IIoT systems. The suggested approach presents the issue as a Multi-Objective Markov Decision Process and adopts an actor-critic learning approach combined with hierarchical and federated learning models. Experimental results with DeepMIMO dataset illustrate that the system has improved performance with throughput of 98.6 Mbps, 10.8 ms latency, 148 J energy consumption and 98.2 packet delivery ratio, which is better than current approaches. The findings validate the usefulness of the proposed framework in the realization of ideal trade-offs between various objectives, which allows efficient, scalable, and intelligent control of networks.