AI-Driven Multi-Objective Optimization for Converged Private 5G and Wi-Fi 7 Industrial Networks

Abstract

Industrial wireless networks are increasingly expected to support workloads that were traditionally confined to wired infrastructure, including real-time robotics, machine vision, and autonomous guided vehicles. The convergence of private 5G and Wi- Fi 7 (IEEE 802.11be) offers a promising path toward this goal, but it also introduces a complex and tightly coupled control space that is difficult to manage using static rules or isolated optimization strategies. This paper presents a practical, AI-driven, multi-objective optimization framework that combines reinforcement learning for radio and slice-level resource control, deep learning for encrypted traffic characterization, and anomaly detection for predictive network assurance. We describe a cloud–edge architecture and a digital twin-based evaluation environment that allow policies to be trained, tested, and deployed under representative industrial workloads. Experimental results show consistent reductions in tail latency and energy consumption, along with improved service-level agreement (SLA) compliance, when compared with conventional rule-based approaches.