Detecting Cellphone Camera Status at Distance by Exploiting Electromagnetic Emanations

Abstract

This paper investigates unintended radiated emissions from cellphones to identify operational status of rear/front camera. We implement a supervised learning method to achieve our goal. In the training phase, we collect data for possible combinations of phone model and camera status. Then, we apply two-phase-dimension-reduction method for better and effective classification. The first dimension-reduction phase is averaging magnitudes of frequency components of a sliding window, which is followed by applying principle component analysis (PCA) technique to reduce the dimension further. In testing phase, k-Nearest-Neighbors (k-NN) algorithm is utilized to classify test data. Finally, we provide examples to show that emanated EM signals from cellphone cameras can exfiltrate useful information.