Machine Learning-Based System Electromagnetic Environment Anomaly Detection Method

摘要

Abnormal electromagnetic signals refer to the insertion of a malicious module in hardware. When a malicious module exchanges information through the self-built channel and the outside world, it has the authority to access all hardware devices, and the threat is huge. In order to effectively identify abnormal electromagnetic signals, we have combined the big data platform technology and machine learning classification technology to propose anomaly detection of electromagnetic signals at the physical layer to find malicious anomalous electromagnetic signals in the hardware. The results show that our method can detect abnormal electromagnetic signals very well, and can reach 98% in the recognition rate of abnormal electromagnetic signals. It has considerable reference value for electromagnetic signal monitoring and network anomaly detection.