Smart user identification using cardiopulmonary activity

In this paper, we present a novel non-intrusive second-factor user authentication mechanism for smart spaces called Cardiopulmonary-ID (alluded as CP-ID), that does not require any special hardware to conform the second factor of authentication. CP-ID uses the unique perturbations in the ambient WiFi signals caused by the cardiopulmonary activity of an individual (referred to as CP-imprint) to confirm the inherence (i.e., something that the user is) as a second-factor of authentication. We show that the CP-imprint of Alice is uniquely manifested in the Channel State Information (CSI) of ubiquitous WiFi signals. We also demonstrate the distinctiveness between the CP-imprints of Alice and Bob. CP-ID proposes a comprehensive pipeline that assists in obtaining a noise-free smooth CP-imprint that is manifested in the CSI of commercial WiFi devices. For feature engineering, CP-ID employs PCA and then formulates a comprehensive set of features extracted from the selected principal components. A multiclass SVM is proposed to identify a target individual from a set of N people. We implement CP-ID using off-the-shelf 802.11n devices and evaluate its performance. Our tests reveal that CP-ID can identify an individual with an average accuracy of 84% to 65% from a group of size 2 to 5 individuals, respectively.