Temporal-Spectral Measuring, Modeling, and Predicting for Industrial EMI

The rise of 5G in industrial environments faces a critical hurdle: pervasive electromagnetic interference (EMI) that threatens the reliability and performance of operational technology (OT) networks. Current research relies on statistical analysis, which struggles to meet industry-standard quality of service (QoS) requirements in the adversarial radio environments. To address this deficiency, based on the on-site EMI measurement, we propose an innovative equipment-specific time domain modeling approach, integrating control instructions for unprecedented granularity in EMI characterization. This methodology complements traditional statistical analysis, enabling a more comprehensive understanding of EMI’s impact. Furthermore, we introduce an EMI prediction algorithm, leveraging both measured data and control instructions, achieving a median EMI power predicting error of less than 0.7 dBW, outperforming the baseline by 5.5 times. Our empirically validated algorithm offers a powerful tool for mitigating EMI’s impact on industrial wireless communication, paving the way for reliable transition toward wireless OT networks, and supporting robust industrial cognitive radio.