Yukio Itakura

Proposal on a multifactor biometric authentication method based on cryptosystem keys containing biometric signatures

Conventional biometric authentication methods have a serious problem – it is easy to steal the biological information of another person, such as fingerprints, iris patterns, and facial forms. Accordingly, it is essential to protect authentication systems from an attack like impersonation with an artificial finger. To solve this problem, in this paper, we would like to propose a multifactor biometric authentication method based on cryptosystem keys containing biometric signatures.This paper has two new aspects: one is to propose a method for embedding two kinds of data – personal biological information and a confidential item corresponding to a conventional secret key – into a cryptosystem key to detect any impersonation even if the biological information is stolen, and the other is to demonstrate the safety of the method, a factor which has never been referenced.From a privacy protection point of view, we believe that the multifactor biometric authentication method is of great significance. Moreover, it can economically establish certification authorities having the biological information. Meanwhile, in order to uniquely embed the biological information into the cryptosystem key, an algorithm is needed to uniquely select a representative template as described in this paper. In this paper, we propose only the concept of this algorithm and hope that concrete methodologies will be studied in the future.

Proposal on personal identifiers generated from the STR information of DNA

Abstract.The individual differences in the repeat count of several bases, short tandem repeat (STR), among all of the deoxyribonucleic acid (DNA) base sequences, can be used as unique DNA information for a personal identification (ID). We propose a method to generate a personal identifier (hereafter referred to as a “DNA personal ID”) by specifying multiple STR locations (called “loci”) and then sequencing the repeat count information. We also conducted a validation experiment to verify the proposed principle based on actual DNA data.We verified that the matching probability of DNA personal IDs becomes exponentially smaller, to about 10-n, as n stages of loci are used and that no correlation exists among the loci.Next, we considered the various issues that will be encountered when applying DNA personal IDs to information security systems, such as biometric personal authentication systems.