AI for Positioning Accuracy Enhancement in the IIoT

Abstract

The evolution of mobile networks, accelerated by the deployment of 5G and the upcoming shift toward 6G, has unlocked new possibilities for industrial applications. A key beneficiary of this progress is the Industrial Internet of Things (IIoT), which depends heavily on precise positioning to support automation, ensure operational safety, and enable real-time system monitoring. Despite these advancements, achieving reliable localization in complex indoor settings remains a persistent challenge for traditional techniques like Time Difference of Arrival (TDoA), which are vulnerable to signal reflections and obstructions caused by multipath propagation. To address these limitations, this study introduces an AI-powered localization method built on a deep learning framework using ResNet architecture. By integrating and analyzing data from multiple sources, the proposed solution overcomes the inherent weaknesses of conventional approaches, delivering enhanced accuracy and resilience in densely packed industrial environments. Comprehensive simulations in realistic indoor factory settings confirm the superiority of the AI-driven model over TDoA, with notable improvements in positioning accuracy. These results underscore the promise of deep learning for advancing IIoT localization, marking a significant step toward intelligent positioning systems in future 5G and 6G networks.