Rui Liu

Topic

Secure and Privacy-preserving Data Aggregation in Internet of Vehicles

Department of Computer Science

Date & location

• Monday, March 4, 2024

• 10:00 A.M.

• Virtual Defence

Reviewers

Supervisory Committee

• Dr. Jianping Pan, Department of Computer Science, University of Victoria (Supervisor)

• Dr. Bruce Kapron, Department of Computer Science, UVic (Member)

• Dr. Issa Traore, Department of Electrical and Computer Engineering, UVic (Outside Member) 

External Examiner

• Dr. Rongxing Lu, Department of Computer Science, University of New Brunswick 

Chair of Oral Examination

• Dr. Daniela Constantinescu, Department of Mechanical Engineering, UVic

   

Abstract

In Internet of Vehicles (IoV), crucial data is aggregated to support the applications for automatic driving, intelligent transportation and smart cities. It is crucial to carefully address certain challenges in this process, particularly regarding security and privacy.

In this thesis, we first target a representative IoV data aggregation scenario, fine grained air quality monitoring. The major challenges we focus on include: a) the sensory data provided by vehicles usually vary in quality; b) there is a significant difference in traffic volumes of streets or blocks, which leads to a data sparsity problem; and c) the original sensory data, vehicle identities, and trajectories face risks of expo sure. To address these issues, we propose a truth discovery algorithm incorporating multiple correlations, and extend it to a privacy-preserving framework, EAirQ.

EAirQ relies on a traditional end-to-end data aggregation architecture. Designing a new architecture specifically for vehicular networks may hold significant value. Thus, we introduce a privacy-preserving two-layered architecture with vehicle clusters. Instead of focusing on a specific application, we present how this architecture can be well adopted in a general distributed machine learning scenario. We named this part of the work CRS. CRS not only protects the local data, the identities and trajectories of vehicles, but also ensures the accuracy of aggregated learning models by handling packet loss in the application layer.

We further work on eliminating the limitations of the proposed two-layered architecture in the following three aspects: a) to provide fast and easy verification of messages within a cluster; b) to preserve vehicle privacy without adopting the pseudonym technique; c) to consider the adversarial behaviors of vehicles and enhance the security. Our solution introduces a novel concept, data approval, based on the Schnorr signature scheme. This part of the work, named as SADA, meets more security requirements and is lightweight for vehicles.

In addition to exploring new solutions to preserve the privacy of vehicle identities and trajectories, we also pay attention to the latest industry standards. This part of the work focuses on tackling the challenge of certificate provisioning in the latest solution to satisfy the anonymous communication requirement in IoV. We propose a non-interactive approach, named as NOINS, empowering vehicles to generate short term key pairs and anonymous implicit certificates on their side. This new paradigm introduces the possibilities for many extensions and applications.