Master Thesis

Biometric authentication is a promising solution for secure authentication. However, the storage of biometric data may lead to security and privacy problems. A Helper Data System (HDS) is introduced to solve these problems. HDS offers the capability to extract information from noisy data while maintaining the confidentiality. There are two types of HDS, which are Fuzzy Extractor (FE) and Secure Sketch (SS). The difference between FE and SS is, a Secure Sketch allows reconstruction of the enrolled biometric from a noisy measurement, whereas a Fuzzy Extractor derives a stable uniform key or secret. In this thesis, we consider a general Helper Data System (HDS), which extracts secret from the enrollment data, but with non-uniform probability of secret. This thesis explore the optimal secret reconstruction procedure in Zero Leakage Helper Data System (ZLHDS). The main goal is to maximize the amount of information that can be extracted from noisy data. The optimization of ZLHDS is done by maximizing the mutual information of the generated secret and the reconstruction secret given helper data. The optimization of ZLSS consists of two phase. First is the analysis to formulate the optimum reconstruction boundary. Second is the numerical analysis to find the optimum probability to obtain the maximum mutual information. The result confirmed that ZLHDS gives a higher mutual information than Zero Leakage Fuzzy Extractor (ZLFE) for a certain amount of noise. This thesis also explored the theoretical approach for the optimization of ZLHDS in a more general case, which is for arbitrary source and arbitrary noise distribution. This approach was done using Lagrange Multiplication, which resulted in a more complicated formula. Optimization of Zero Leakage Helper Data Systems iii Acknowledgement I would like to express my sincere gratitude to my supervisor dr. Boris Skoric for the continuous support of my thesis study and research, for his patience, motivation, enthusiasm, and immense knowledge. Besides my supervisor, I would like to thank the rest of my thesis committee: dr.ir. L.A.M. (Berry) Schoenmakers and dr. Tanya Ignatenko for their cooperation, insightful comments, and hard questions. iv Optimization of Zero Leakage Helper Data Systems