Differential privacy for location and trajectory data

Details

Supervisors

Pereira, Olivier ; Standaert, François-Xavier

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en mathématiques appliquées, à finalité spécialisée

Abstract

As data-driven technologies continue to shape our world, the need for privacy and security in the realm of data has become increasingly important. The proliferation of services that rely on personal data has created a massive playground for cyber attacks. A specific area where highly sensitive data is processed is Location-based services. Location-based services (LBS) are mobile software services that utilize a device's location data, typically acquired through GPS or cellular triangulation, to offer users relevant information or services based on their current or past locations. Examples of LBS include navigation and mapping applications, ride-sharing services, weather and news apps, and social media platforms that offer location-based features such as check-ins and location tagging. But location data has applications beyond Location-based services. Location data is a highly valuable asset that can be used in a multitude of ways. It can be useful for a variety of purposes, including: - Providing personalized services: Location data can be used to offer personalized services to users. For example, a ride-sharing app can use location data to match riders with nearby drivers, or a restaurant recommendation app can suggest nearby dining options based on the user's location. - Analyzing user behavior: Location data can be used to analyze user behavior and preferences, which can help companies better understand their target audience and improve their products or services. For example, a retailer can use location data to track foot traffic in their stores and optimize store layouts and product placements accordingly. - Enhancing public safety: Location data can be used to improve public safety by helping emergency responders locate individuals in need of assistance. For example, Police dispatchers can use location data to pinpoint the location of a caller and send help to their exact location. - Enabling location-based advertising: Location data can be used to deliver targeted advertisements to users. For example, a retailer can use location data to offer promotions to users when they are near their stores. With the spread of location-based services and the resulting increase in circulation of location data, a lot of research has been conducting on quantifying, analyzing the risks of de-anonymization [2,24,35] and implementing solutions to enhance the level of privacy in location data publishing. Some of these researches have been conducted on the subject of differential privacy and how it can be suitable for location data [3,8,10,28]. Differential privacy, as presenter by Dwork [40] provides a mathematical definition and methodology for adding noise or randomness to data queries or releases, thereby protecting individual privacy and preventing the disclosure of sensitive information. This thesis aims to address the privacy challenges associated with location-based services through two distinct aspects. A clear distinction is made between location data and trajectory data. The examination of these data types needs to be conducted separately due to the disparity between the utility that can be derived from each type of data and the relevance of privacy-preserving tools for these two data categories. Firstly, the thesis will examine location data from two different perspectives. We will explore the scenario where an application utilizes location data within a confined space, followed by the scenario where an application utilizes outdoor location data. In each case, the utility of the application will be clearly defined, and a privacy-preserving mechanism will be presented accordingly. Secondly, we will address the sensitivity of trajectory data within a running application. Here, once again, it will be essential to clearly define the utility of such an application, and then we will strive to develop a privacy-preserving solution tailored to this specific context.