Reichert, Leonie (2024)
Privacy-Preserving Data Analysis and Distributed Processing in Pandemic Settings and Beyond.
Technische Universität Darmstadt
doi: 10.26083/tuprints-00028595
Dissertation, Erstveröffentlichung, Verlagsversion
Kurzbeschreibung (Abstract)
Privacy is acknowledged as a fundamental human right and essential for the functioning of modern democracies, particularly as research and the economy become increasingly data driven. The Covid-19 pandemic has given rise to many new applications necessitating the processing of sensitive information such as health, location, and proximity data. Notable examples include discovering new infections by retracing the contacts of diagnosed individuals and identifying super-spreader events through presence tracing. To fight a pandemic, gaining meaningful statistical insight on the current epidemiological situation based on health data - or other sensitive information - is important. For all these applications, the processed data can reveal private information regarding the data providers. Therefore, data providers require concrete privacy guarantees at every step.
This thesis focuses on solutions for processing and analyzing sensitive data in a privacy-preserving way without requiring trust in a central authority. Multiple approaches are proposed to ensure privacy during distributed data processing for Digital Contact Tracing (DCT). To establish a general understanding of the topic, an introduction to the problems and solutions for DCT is presented. The literature is systematized and common challenges with regard to privacy, security, and functionality are identified. Based on the shortcomings of existing contact tracing applications, novel designs for privacy-preserving DCT are presented, along with their respective advantages and drawbacks. The focus is on distributing the tracing process and risk-scoring tasks to users while mitigating the leakage of private data through metadata. Strong privacy guarantees are also provided by using cryptographic primitives such as blind signatures, Oblivious Random Access Memory (ORAM), and Private Set Intersection (PSI). Such techniques allow the design of protocols that only reveal the minimal required amount of information to all parties involved. Systems for super-spreader detection through presence tracing are also presented that can be integrated with DCT systems in a privacy-preserving manner.
While decentralized processing provides better privacy than the centralized alternative, it limits the ability to observe the epidemic situation through statistical analysis. By reviewing common approaches for collecting and analyzing health data for research purposes, we identify various threats to the privacy of people who are willing to share their data. Both in the pandemic and post-pandemic settings, privacy guarantees are a tool to ensure to data providers that their data can not be misused. To this end, a platform is presented that leverages Trusted Execution Environments (TEEs) in combination with oblivious algorithms that safeguard sensitive data during data collection and analysis. To combat the drawbacks of TEEs, new methods are introduced to hide the access patterns and volume patterns of database queries. All contributions presented in this thesis aim to improve the privacy of individuals through solutions that follow the concept of privacy by design.
Typ des Eintrags: | Dissertation | ||||
---|---|---|---|---|---|
Erschienen: | 2024 | ||||
Autor(en): | Reichert, Leonie | ||||
Art des Eintrags: | Erstveröffentlichung | ||||
Titel: | Privacy-Preserving Data Analysis and Distributed Processing in Pandemic Settings and Beyond | ||||
Sprache: | Englisch | ||||
Referenten: | Scheuermann, Prof. Dr. Björn ; Lueks, Dr. Wouter | ||||
Publikationsjahr: | 28 November 2024 | ||||
Ort: | Darmstadt | ||||
Kollation: | 14, 211 Seiten | ||||
Datum der mündlichen Prüfung: | 8 Oktober 2024 | ||||
DOI: | 10.26083/tuprints-00028595 | ||||
URL / URN: | https://tuprints.ulb.tu-darmstadt.de/28595 | ||||
Kurzbeschreibung (Abstract): | Privacy is acknowledged as a fundamental human right and essential for the functioning of modern democracies, particularly as research and the economy become increasingly data driven. The Covid-19 pandemic has given rise to many new applications necessitating the processing of sensitive information such as health, location, and proximity data. Notable examples include discovering new infections by retracing the contacts of diagnosed individuals and identifying super-spreader events through presence tracing. To fight a pandemic, gaining meaningful statistical insight on the current epidemiological situation based on health data - or other sensitive information - is important. For all these applications, the processed data can reveal private information regarding the data providers. Therefore, data providers require concrete privacy guarantees at every step. This thesis focuses on solutions for processing and analyzing sensitive data in a privacy-preserving way without requiring trust in a central authority. Multiple approaches are proposed to ensure privacy during distributed data processing for Digital Contact Tracing (DCT). To establish a general understanding of the topic, an introduction to the problems and solutions for DCT is presented. The literature is systematized and common challenges with regard to privacy, security, and functionality are identified. Based on the shortcomings of existing contact tracing applications, novel designs for privacy-preserving DCT are presented, along with their respective advantages and drawbacks. The focus is on distributing the tracing process and risk-scoring tasks to users while mitigating the leakage of private data through metadata. Strong privacy guarantees are also provided by using cryptographic primitives such as blind signatures, Oblivious Random Access Memory (ORAM), and Private Set Intersection (PSI). Such techniques allow the design of protocols that only reveal the minimal required amount of information to all parties involved. Systems for super-spreader detection through presence tracing are also presented that can be integrated with DCT systems in a privacy-preserving manner. While decentralized processing provides better privacy than the centralized alternative, it limits the ability to observe the epidemic situation through statistical analysis. By reviewing common approaches for collecting and analyzing health data for research purposes, we identify various threats to the privacy of people who are willing to share their data. Both in the pandemic and post-pandemic settings, privacy guarantees are a tool to ensure to data providers that their data can not be misused. To this end, a platform is presented that leverages Trusted Execution Environments (TEEs) in combination with oblivious algorithms that safeguard sensitive data during data collection and analysis. To combat the drawbacks of TEEs, new methods are introduced to hide the access patterns and volume patterns of database queries. All contributions presented in this thesis aim to improve the privacy of individuals through solutions that follow the concept of privacy by design. |
||||
Alternatives oder übersetztes Abstract: |
|
||||
Freie Schlagworte: | Privacy by Design, Automated Contact Tracing, Digital Health, Covid-19, Pandemic Measures, Data Donations, Privacy-Preserving Data Sharing | ||||
Status: | Verlagsversion | ||||
URN: | urn:nbn:de:tuda-tuprints-285959 | ||||
Sachgruppe der Dewey Dezimalklassifikatin (DDC): | 000 Allgemeines, Informatik, Informationswissenschaft > 004 Informatik 600 Technik, Medizin, angewandte Wissenschaften > 600 Technik 600 Technik, Medizin, angewandte Wissenschaften > 610 Medizin, Gesundheit |
||||
Fachbereich(e)/-gebiet(e): | 18 Fachbereich Elektrotechnik und Informationstechnik 18 Fachbereich Elektrotechnik und Informationstechnik > Institut für Datentechnik 18 Fachbereich Elektrotechnik und Informationstechnik > Institut für Datentechnik > Kommunikationsnetze |
||||
Hinterlegungsdatum: | 28 Nov 2024 14:31 | ||||
Letzte Änderung: | 02 Dez 2024 08:50 | ||||
PPN: | |||||
Referenten: | Scheuermann, Prof. Dr. Björn ; Lueks, Dr. Wouter | ||||
Datum der mündlichen Prüfung / Verteidigung / mdl. Prüfung: | 8 Oktober 2024 | ||||
Export: | |||||
Suche nach Titel in: | TUfind oder in Google |
Frage zum Eintrag |
Optionen (nur für Redakteure)
Redaktionelle Details anzeigen |