Field Deviation in Radiated Emission Measurement in Anechoic Chamber for Frequencies up to 60 GHz and Deep Learning-Based Correction

Abstract

The measurement of radiated emission (RE) in an anechoic chamber becomes very challenging at high frequencies, up to 60 GHz, because the scanning plane of the receiver is in measurement standard deviation from the actual wavefront. As a result, the RE intensity of the devices may be underestimated, resulting in electromagnetic interference. The deviation between the electric field at the far-field vertical scanning point and the actual wavefront is researched. Then, in an anechoic chamber, a hybrid deep learning amendment model of convolutional neural network (CNN) and transformer is proposed to correct the RE measurement at a 3 m distance. The results indicate that the correction is reliable, with an average error of 6.35% for a 3 m distance in a semianechoic chamber and less than 4.83% for other test scenarios. The proposed method provides a promising solution for RE measurement at a millimeter wave band in an anechoic chamber.