Deep Learning-Based Estimation of Impacts of Root Causes on 5G Throughput Performance

It is becoming increasingly important to fully utilize the benefits of wireless technology such as 5G to improve productivity by utilizing industrial Internet of Things. To protect productivity from the root causes of radio-quality degradation, such as distance attenuation, shadowing, and fading, it is necessary to automatically identify these root causes and quickly take measures to address each cause. However, since these root causes occur simultaneously, it is necessary to have an expert verify which causes are affecting communication quality and to what extent. This paper proposes a method that combines learning and simulation to estimate how much communication quality has degraded for each root cause. Through a simulation evaluation, we confirmed that the proposed method can estimate the degree of throughput degradation with 90% accuracy.