Abstract
We investigate the security of a privacy enhancing technique for ngerprint authentication known as fuzzy ngerprint vault. This technique is based on the fuzzy vault of Jules and Sudan, a scheme that allows error tolerant authentication, while preserving the privacy of the reference data. We explore if and under what circumstances a secure fuzzy fingerprint vault can be implemented. We derive both upper and lower security bounds for any attacks that attempt to recover the template from the stored reference data, and, at the same time, signicantly improve the best known attack. Furthermore, we show how to select optimal parameters and evaluate both minimum minutiae match rates and minimum number of minutiae needed to obtain an appropriate security level. Our results quantify the security capacity of the fuzzy fingerprint vault and provide important tools for selection of suitable parameters.