An efficient protocol for private iris-code matching by means of garbled circuits

Privacy Protection in Biometric-Based Recognition Systems: A marriage between cryptography and signal processing

“…In [31] and [10], the authors present an iris identification protocol based on two different full-GC implementations (more details are given in the next section). In [31], the authors run a Java implementation of the protocol on a client with a 2.66-GHz quad-core processor connected through a local area network with a server equipped with a 2-GHz processor.…”

“…The protocol described in [10] has been implemented in Java and run on a machine mounting a 3.00-GHz processor over IrisCodes of the CASIA Iris database [29] represented with 9600 and 2,048 bits. Thanks to offline computation of the circuit garbling phase and circuit transmission, the matching between two IrisCodes represented with 2,048 bits needs 0.56 seconds and the transmission of 571 kilobytes, while the matching between two IrisCodes represented with 9,600 bits needs 2.5 seconds and the transmission of 2,655 kilobytes.…”

“…It is then necessary to find a proper configuration to couple efficiency and accuracy. Signal processing expertise can be exploited in several other ways: for example, it has been proven in [10] that using a common mask for iris recognition instead of a varying one dramatically simplifies the implementation of an iris-recognition system in the encrypted domain, with a very reduced impact on the performance of the system. This article aims to illustrate the basics of STPC, including the way it can be applied to the protection of biometric templates, and to explain how the signal processing and cryptographic points of view can be considered together to obtain efficient, secure, and accurate SPED protocols.…”

Secure Evaluation of Private Functions through Piecewise Linear Approximation

“…Recently, Fully Homomorphic Encryption (FHE) schemes [6] have been proposed, permitting to evaluate any Boolean function without interaction, however FHE techniques are still inefficient, due to the bit-length of ciphertexts (each one encrypting a single bit) and public key. SMPC has been applied to different areas wherein cooperation between parties faces with privacy issues: biometry matching [7], [8], [9], remote helthcare analysis [10], etc. In these scenarios, SMPC is used to protect the involved values, however the evaluated functionality is public, allowing a service provider to protect the inputs and parameters of the functionality, but not the functionality itself (or part of it).…”

Secure multi-party consensus gossip algorithms

“…As well as in other privacy preserving applications such as data mining [15], biometric matching [16], recommendation systems [17], biomedical analysis [18], etc., the consensus protocol can be executed in the encrypted domain [6] by using secure Multi-Party Computation (MPC) protocols to reach the consensus about a common value while each agent has access only to its inputs and the final decision, obtained by evaluating the protocol on encrypted statistics.…”