A USB Keyboard Electromagnetic Information Detection Algorithm Based on Deep Learning

The keyboard is a commonly used computer input device, which inevitably generates electromagnetic leakage during the process of key input, resulting in the risk of information leak-age. This article analyzes the working principle and signal characteristics of USB keyboard EM INFO security issues, and proposes a deep learning-based USB keyboard EM leakage detection algorithm. The algorithm adaptively adjusts the convolutional neural network structure based on the characteristics of keyboard EM leakage signals, introduces gating recursive units and gradient weighted activation mapping units, and combines sample preprocessing based on end-point detection to recognize and locate the EM signals of 39 commonly used keyboard keys.