Deep Learning-Based Positioning System for Underground MIMO Communications: A CNN Approach for Complex Tunnel Environments

Abstract

Underground mining operations demand robust communication and positioning systems to ensure safety and operational efficiency. However, the distinctive characteristics of these environments, such as multipath signal propagation and the unavailability of satellite-based navigation systems, present significant challenges. Traditional positioning methods often rely on LOS signal processing, which can be limited in such complex settings. To address these limitations, this paper introduces a convolutional neural network (CNN)-based approach for position identification in MIMO channels, utilizing machine learning techniques to enhance accuracy and reliability in underground environments. Validation using real-world measurement data demonstrates the effectiveness of this approach, achieving high identification accuracy and highlighting its potential to improve positioning systems in these challenging conditions.