Monitor Identification Based on Electromagnetic Emanations and Multilayer Perceptron Networks

We present an approach for automatic identification of type and state (on/off) of computer monitors from EM emanations measured in the presence of multiple other monitors that create electromagnetic noise. The proposed approach is based on multilayer perceptron neural networks trained on measured electromagnetic field. We use ensemble of 100 networks for automatic selection of significant spectral subranges of measured emanations.