A DETECTION METHOD FOR THE SUBLIMINAL CHANNEL UNDER THE EDGE COMPUTING

摘要

The invention of A detection method for the subliminal channel under the edge computing under the edge computing is published. This method mainly includes three steps. Firstly, Mel-frequency cepstral coefficients (MFCCs) features are generated; secondly, the subliminal channel identification system is trained to utilize machine learning method to classify the behavior of normal communication system or subliminal channel communication system; thirdly, the identification system is verified by inputting MFCCs features of specific equipment into identification system to test the recognition effect of the covert communication behavior of the submarine channel; The edge computing platform, as the control node of the edge network, collects the MFCCs features of the terminal, and conducts offline training and online decision-making. After that, the edge computing platform can identify whether a subliminal channel communication behavior has been issued while the edge computing platform communicates with the terminals. Besides, the advantages of this method are low computational complexity and high recognition accuracy.