Data Privacy for Big Automotive Data

Licentiate thesis, 2017

In an age where data is becoming increasingly more valuable as it allows for data analysis and machine learning, big data has become a hot topic. With big data processing, analyses can be carried out on huge amounts of user data. Although big data analysis has increased the ability to learn more about a population, it also carries a risk to individual users’ privacy, as big data can contain or reveal unintended personal information. With the growing capacity to store and process such big data, the need to provide meaningful privacy guarantees to users thus becomes a pressing issue. We believe that techniques for privacy-preserving data analysis en- ables big data analysis, by minimizing the privacy risk for individuals. In this work we have further explored how big data analysis can be enabled through privacy-preserving techniques, and what challenges arise when implementing such analyses in a real setting. Our main focus is on differential privacy, a privacy model which protects individuals’ privacy, while still allowing analysts to learn sta- tistical information about a population. In order to have access to real world use cases, we have studied privacy-preserving big data analysis in the context of the automotive domain.