Fundamental Limits of Identification System With Secret Binding Under Noisy Enrollment
share
We investigate fundamental limits among identification, secrecy, template, and privacy-leakage rates in biometric identification system with secret binding from an information theoretic perspective. Ignatenko and Willems (2015) studied the system with estimation of individuals and clarified the fundamental limit among identification, secrecy, and privacy-leakage rates provided that the enrollment channel of the system is noiseless. In the enrollment process, however, it is highly considerable that noise occurs when bio-data is scanned. Recently, Günlü and Kramer (2018) analyzed the system assuming that the channel is noisy when there is only one individual. In this paper, we analyze the model considering the estimation of individuals under noisy enrollment and characterize the capacity region of identification, secrecy, template, and privacy-leakage rates in the biometric identification system with secret binding. The obtained result shows that it reduces to the one given by Ignatenko and Willems (2015) as a special case where the enrollment channel is noiseless.
READ FULL TEXT