Privacy Preserving Control Using Multiparty Computation: Security
Author
Mølgaard, Katrine Tjell
Term
4. semester
Education
Publication year
2018
Submitted on
2018-06-07
Pages
96
Abstract
Flere og flere enheder bliver smarte og kan selv træffe beslutninger. Et smarthjem kan for eksempel regulere varmen, så der er behageligt, når beboerne er hjemme, og spare energi, når huset er tomt, eller endda låse op, når beboere nærmer sig døren. For at træffe sådanne valg har systemet brug for personlige data, som mange ikke ønsker at dele. Derfor er der behov for privatlivsbevarende metoder, der kan lave beregninger på data, mens de holdes skjult med kryptering. Denne afhandling undersøger potentialet i at bruge sikker flerpartsberegning (secure multiparty computation, MPC) til at skabe sådanne metoder. MPC gør det muligt for flere parter at beregne et fælles resultat uden at afsløre deres individuelle input. Afhandlingen finder, at velkendte algoritmer som gradient descent-metoden og de rekursive mindste kvadraters ligninger kan formuleres som sikre flerpartsprotokoller. Det peger på, at tilgangen er gennemførlig og giver grundlag for videre forskning.
More and more devices are becoming smart and can make their own decisions. A smart home, for example, could adjust heating so it is comfortable when people are home and saves energy when the house is empty, or even unlock when residents approach the door. To make such choices, the system needs personal data, which many people are reluctant to share. This creates a need for privacy-preserving methods that can perform computations on data while it remains hidden by encryption. This thesis investigates the potential of using secure multiparty computation (MPC) to build such methods. MPC allows several parties to compute a joint result without revealing their individual inputs. The thesis finds that well-known algorithms, such as the gradient descent method and the recursive least squares equations, can be formulated as secure multiparty protocols. This indicates that the approach is feasible and supports further research in this area.
[This abstract was generated with the help of AI]
Keywords
Documents