The rapid development of information technology not only leads to great convenience in our daily lives, but also raises significant concerns in the field of security and privacy. Particularly, the authentication process, which serves as the first line of information security by verifying the identity of a person or device, has become increasingly critical. An unauthorized access could result in detrimental impact on both corporation and individual in both secrecy loss and privacy leakage. Unlike many existing studies on user/device authentication, which either employ specialized or expensive hardware that needs experts for installation and calibration or require users' active involvement, the emerging low-cost and unobtrusive authentication solution without the users' participation is particularly attractive to effectively complement conventional security approaches. Due to the rich wireless connectivity and unique signal characteristics in pervasive wireless environments, this project takes a different view point by exploiting unique physical properties in wireless networks to facilitate implicit authentication for both human and mobile devices. The proposed research could advance our knowledge in exploiting the physical layer information in wireless networks to capture unique physiological and behavioral characteristics from human during their daily activities. It could also enhance our understanding in developing deep learning techniques to authenticate people based on their activities in the physical environments. Additionally, the educational efforts include curriculum development, K-12 and undergraduate involvement, and underrepresented student engagement in research.This project focuses on building a holistic framework that leverages fine-grained radio signals available from the commercial wireless networks to perform implicit user/device authentication. The proposed framework aims to advance the foundation of integrating fine-grained physical properties in wireless networks to enhance wireless security. The research reveals that the fine-grained signal properties in wireless networks are capable to capture unique physiological and behavioral characteristics from human in both stationary and mobile daily activities. The proposed framework develops smart segmentation on the wireless signals and extract unique features that enable the capability of distinguishing individual. It further develops deep learning techniques to authenticate people based on their daily activities in the physical environments. The authentication process does not require active user involvement nor require the user to wear any device. This project also develops efficient techniques to detect the presence of user spoofing and localize attackers to facilitate the employment of a broad array of defending strategies.
base; Behavioral; Calibration; Characteristics; Complement; cost; curriculum development; Data Security; deep learning; Development; Devices; Employment; Environment; Extravasation; Foundations; Grain; handheld mobile device; Human; Human Characteristics; Individual; Information Technology; Knowledge; Names; Persons; Physical environment; physical property; Physiological; Privacy; Process; Property; Radio; Research; Secrecy; Security; Signal Transduction; student participation; Techniques; undergraduate student; Underrepresented Students; wireless network; Wireless Technology