NEC streamlines 1:N biometric matches for homomorphic encryption to protect data
Homomorphic encryption for biometric matching holds the promise of data protection even in use, and NEC Corporation is the latest technology provider to develop a system that it says delivers on this promise, with the key difference that the company says it can be used for one-to-many searches.
Biometric data encrypted between collection and transmission to a server or service provider for matching prevents the leakage of raw images, which can be subsequently utilized in spoofing attacks. In the system developed by NEC, the decryption key is held by the user, rather than the service provider, providing users with additional assurance their information is protected.
Homomorphic encryption can be used for authentication without reducing accuracy, according to the announcement, but only in simple operations, with processing speed slowing down significantly as soon as any complexity in introduced. The technology has therefore traditionally been limited to 1:1 biometric matching for identity verification.
NEC’s new technology streamlines the processing of homomorphic encryption to enable high-speed 1:N identification, the company claims.
The new technology can narrow down user candidates from a biometric database of 10,000 identities in 0.01 seconds, and if it reaches 1 percent, or 100 candidates, perform face authentication in about a second, with no impact on accuracy.
NEC plans to integrate the capability with its Bio-IDiom portfolio of authentication technologies.
Homomorphic encryption was identified as a likely area of increased adoption in the year ahead in a recent forecast from Deloitte Global.