Ph.D. Defence Notice - Cheng Huang

Wednesday, August 12, 2020 10:00 am - 10:00 am EDT
Candidate: Cheng Huang
Title: Effective Privacy-preserving Mechanisms for V2X Services
Date: August 12, 2020
Time: 10:00 AM
Supervisor(s): Shen, Sherman

Owing to the advancement of wireless communication technologies, drivers can rely on smart connected vehicles to communicate with each other, roadside units, pedestrians, and remote service providers to enjoy a large amount of vehicle-to-everything (V2X) services, including navigation, parking, ride hailing, and car sharing. These V2X services provide different functions for bettering travel experiences, which have a bunch of benefits. In the real world, even without smart connected vehicles, drivers as users can utilize their smartphones and mobile applications to access V2X services and connect their smartphones to vehicles through some interfaces, e.g., IOS Carplay and Android Auto. In this way, they can still enjoy V2X services through modern car infotainment systems installed on vehicles.

Most of the V2X services are data-centric and data-intensive, i.e., users have to upload personal data to a remote service provider, and the service provider can continuously collect a user’s data and offer personalized services. However, the data acquired from users may include users’ sensitive information, which may expose user privacy and cause serious consequences. To protect user privacy, a basic privacy-preserving mechanism, i.e., anonymization, can be applied in V2X services. However, a big obstacle arises as well: user anonymization may affect V2X services’ availability. As users become anonymous, users may behave selfishly and maliciously to break the functions of a V2X service without being detected and the service may become unavailable. In short, there exist a conflict between privacy and availability, which is caused by different requirements of users and service providers. In this thesis, we have identified three major conflicts between privacy and availability for V2X services: privacy vs. linkability, privacy vs. accountability, privacy vs. reliability, and then have proposed and designed three privacy-enhanced mechanisms to resolve these conflicts.

Firstly, the thesis investigates the conflict between privacy and linkability in an automated valet parking (AVP) service, where users can reserve a parking slot for their vehicles such that vehicles can achieve automated valet parking. As an optional privacy-preserving measure, users can choose to anonymize their identities when booking a parking slot for their vehicles. In this way, although user privacy is protected by anonymization, malicious users can repeatedly send parking reservation requests to a parking service provider to make the system unavailable (i.e., “Double-Reservation Attack”). Aiming at this conflict, a security model is given in the thesis to clearly define necessary privacy requirements and potential attacks in an AVP system, and then a privacy-preserving reservation scheme has been proposed based on BBS+ signature and zero-knowledge proof. In the proposed scheme, users can keep anonymous since users only utilize a one-time unlinkable token generated from his/her anonymous credential to achieve parking reservations. In the meantime, by utilizing proxy re-signature, the scheme can also guarantee that one user can only have one token at a time to resist against “Double-Reservation Attack”.

Secondly, the thesis investigates the conflict between privacy and accountability in a car sharing service, where users can conveniently rent a shared car without human intervention. One basic demand for car sharing service is to check the user’s identity to determine his/her validity and want the user to be accountable if he/she did improper behavior. If the service provider allows users to hide their identities and achieve anonymization to protect user privacy, naturally the car sharing service is unavailable. Aiming at this conflict, a decentralized, privacy-preserving, and accountable car sharing architecture has been proposed in the thesis, where multiple dynamic validation servers are employed to build decentralized trust for users. Under this architecture, the thesis proposes a privacy preserving identity management scheme to assist in managing users’ identities in a dynamic manner based on a verifiable secret sharing/redistribution technique, i.e. the validation servers who manage users’ identities are dynamically changed with the time advancing. Moreover, the scheme enables a majority of dynamic validation servers to recover the misbehaving users’ identities and guarantees that honest users’ identities are confidential to achieve privacy preservation and accountability at the same time.

Thirdly, the thesis investigates the conflict between privacy and reliability in a crowdsourcing-based road condition monitoring service, where users can report road conditions to a monitoring service provider to help construct a live map. Usually, a reputation-based mechanism is applied in the service to measure a user’s reliability. However, this mechanism cannot be easily integrated with a privacy-preserving mechanism based on user anonymization. When users are anonymous, they can upload arbitrary reports to destroy the service quality and make the service unavailable. Aiming at this conflict, a privacy-preserving crowdsourcing-based road condition monitoring scheme has been proposed in the thesis. By leveraging homomorphic commitments and PS signature, the scheme supports anonymous user reputation management without the assistance of any third-party authority. Furthermore, the thesis proposes several zero-knowledge proof protocols to ensure that a user can keep anonymous and unlinkable but a monitoring service provider can still judge the reliability of this user’s report through his/her reputation score.

To sum up, with more attention being paid to privacy issues, how to protect user privacy for V2X services becomes more significant. The thesis proposes three effective privacy-preserving mechanisms for V2X services, which resolve the conflict between privacy and availability and can be conveniently integrated into current V2X applications since no trusted third party authority is required. The proposed approaches should be valuable for achieving practical privacy preservation in V2X services.