TUM Logo

Designing privacy-preserving edge services in the automotive domain

Designing privacy-preserving edge services in the automotive domain

Supervisor(s): Christian Banse
Status: finished
Topic: Others
Author: Immanuel Kunz
Submission: 2019-03-15
Type of Thesis: Masterthesis
Thesis topic in co-operation with the Fraunhofer Institute for Applied and Integrated Security AISEC, Garching

Description

Nowadays, evermore sensors and functionalities are built into automobiles generating a multitude of data, e.g. location data, usage data and drive train metrics. Original Equipment Manufacturers (OEMs), component suppliers and other companies want to capture and process this data to be able to provide and improve services like car sharing and dynamic insurance scores. Often this processing is done outside the car in cloud back ends. Since this data can be personal and sensitive, the question arises if and how these services can be designed in a way that preserves the user’s privacy. From the driver’s perspective, it is often unclear which data is being captured and how it is used by OEMs and other service providers. From the service provider’s perspective, there is a lack of standards and widely accepted methodologies about how to select appropriate anonymization techniques and how to design privacy-preserving automotive services. At the same time, service providers and OEMs are tasked with compliance to corresponding regulations, such as the General Data Protection Regulation (GDPR), which entail a significant financial risk.
The thesis’ contribution is twofold. First, it proposes a process for the selection of privacy enhancing technologies (PETs) for automotive data. Second, a framework for the application of PETs is proposed. It provides a context for the execution of PETs and can be deployed close to the user, e.g. on an edge device, to minimize the flow of personal data to cloud back ends and facilitates various other privacy goals.
A prototypical use case based on a car sharing service demonstrates how the framework and the selection process can be applied in practice. The prototype has been implemented using Amazon Web Services, most notably the Greengrass software that allows to deploy edge-based functionality.