Wireless Industrial Channel Characterization Through Machine Learning and Ray Tracing Simulations

This paper investigates the performance of both a machine learning-based wireless propagation model and an empirical formula in an industrial environment. The study examines a configuration with the transmitter placed at the center of the ceiling, considering key propagation parameters, including path loss, shadowing, and delay spread. The machine learning model and the empirical formula applied for predicting propagation parameters demonstrate accurate results. While the empirical formula is simpler to apply, the machine learning model consistently outperforms the empirical approach in terms of predictive accuracy, making it a more effective solution for real-time wireless network optimization. Furthermore, a comparison with previous work, where transmitters were wall-mounted, reveals significant improvements in communication performance with the new configuration, as expected. The previous study was constrained to analyzing a single transmitter height, primarily due to the associated computational effort. That work introduces a new transmitter height into the analysis. The results show remarkable changes and enhancements in the propagation parameters.