Abstract: The fifth generation mobile communication technology uses millimeter wave Multiple-Input Multiple-Output antenna to provide technical support for high quality wireless communication, which brings new challenges to traditional radiation safety assessment. This paper proposes an unsupervised deep learning method to analyze the upper limit of electromagnetic radiation of millimeter wave MIMO equipment, and then evaluate the radiation safety of MIMO equipment. This paper first theoretically analyzes the feasibility of using unsupervised learning methods to analyze the upper limit of electromagnetic radiation of MIMO devices, and takes different antenna arrays with operating frequencies of 28 GHz as examples to verify the feasibility of the proposed method. In addition, the reliability of the proposed algorithm was verified by Monte Carlo method as a reference. The results show that the error of the maximum and mean values of the power density distribution between the proposed method and Monte Carlo method can be controlled below 0. 07%. The proposed method can accurately and efficiently analyze the radiation upper limit of the millimeter wave MIMO equipment, which provides a new scheme for rapid evaluation of the radiation safety of millimeter wave MIMO equipment.

Key words: electromagnetics, unsupervised learning, multiple-input multiple-output antenna