IFace: A Deepfake Resilient Digital Identification Framework for Smart Cities

Abstract

Digital ID is the gateway of “Smart City” for “Smart Citizens”. It gives citizen access to all other stakeholders of smart cities like smart healthcare, smart transport, smart finance, smart energy, etc. effectively and easily. In this paper, we propose a biometric based digital ID which is implemented in IoT environment. It is a secured and robust system against deepfake attacks. A convolutional neural network (CNN) based feature extraction method has been employed to defeat deepfake attacks. The dlib face detector has been used in detecting face landmark points and in calculating distances in the iris and nose region to obtain unique facial features. A bio-key is generated from the combination of features from facial landmarks and various facial distances along with the username. An encoded key is stored in a cloud database during the registration process of the user. For accessing any facilities in a smart city, the user needs to be authenticated. The authentication process is performed at the edge. Small changes in an image due to unconstrained settings are corrected using the Reed Solomon algorithm. Once authenticated at a particular smart facility, the user is now eligible to use that facility.