As cloud-based data has grown increasingly prevalent, data protection has become a critical issue, and homomorphic encryption, which allows mathematical processes to be performed on data in its encrypted state, has gained attention. Existing homomorphic encryption methods, however, use bit-by-bit encryption, which slows down processing times, making practical application of it problematic.
This technology makes it possible to use cloud-based data in more ways while safeguarding privacy. For example, when applied to biometric authentication, such as a fingerprint or vein data, this technology makes it possible to securely match encrypted data without having to decrypt it. Using confidential information such as medical or biological data for data analysis has, up until now, been problematic from a privacy standpoint.
Details of this technology are being presented at the
The increasing prevalence of cloud-based data and mobile devices has led to the emergence of a number of new information services to meet people's needs. At the same time, there is an increasing awareness of the problem of personal information becoming public and of the need to be able to use personal data while keeping it private.
Encryption is an effective way to protect data, although in most encryption methods, data needs to be temporarily decrypted in order to perform calculations such as totals. This is problematic because the data becomes vulnerable the moment it is decrypted. Homomorphic encryption, however, allows for calculations to be performed on data in an encrypted state, making it a promising technology for delivering new cloud services.
Homomorphic encryption, which allows calculations to be performed on encrypted data, has typically encrypted data at the bit-level. Furthermore, when performing statistical calculations between encrypted data, after multiplying each bit in each piece of encrypted data, the results are added to calculate an inner product. This is problematic in that processing times are directly proportional to bit lengths, making practical implementation of this method difficult.
About the Technology
1. Faster batch encryption of multiple bits
When encrypting two plain texts, this takes advantage of the characteristics of polynomial multiplication, reordering the bit string of one in ascending order, the other in descending order, and then converting both to polynomials, which makes it possible for inner products of encrypted bit strings to be calculated as a batch. Compared to the conventional bit-by-bit ciphering process, this results in dramatically faster performance. For example, with 2048 bits of data, the process can be as much as 2048 times as fast. The processing time in proportion to the bit length is greatly reduced.
2. Permits many practical encipher functions
This technology makes it possible to perform numerous useful calculations on enciphered data, such as totals, averages, and standard deviations (which are widely used statistical functions), correlation analysis, and comparisons on biometric data.
This technology makes it possible to use data while safeguarding privacy. For example, an application of this technology that compares biometric data could keep highly secret fingerprints or vein patterns encrypted during the comparison process. With a 2,048-bit feature code(2) extracted from vein-pattern information, that code can be encrypted using homomorphic encryption and the resulting encrypted data can be compared. While the comparison process would take a dozen seconds or more using typical methods, with this technique it can be done in several milliseconds (Figure 4).
Biometric authentication is already being used by banks and corporations for high-security systems, but keeping that biometric information always secure with encryption will make these systems easier to use, as well as open up new areas for short-term identity verification, such as in theme parks, resort hotels, and airport baggage claims, using the human body in place of a key or password (Figure 5).
This technology also opens the door to analysis of highly personal information, such as medical or biological data, by enabling companies to share information while maintaining a wall of privacy.
Glossary and Notes
(1) Homomorphic encryption:An encryption method that allows mathematical operations, such as addition and multiplication, to be performed on encrypted data. Because the results are also encrypted, learning the results requires access to a private key.
(2) 2,048-bit feature code:See "
Fujitsu LimitedPublic and Investor Relations www.fujitsu.com/global/news/contacts/ +81-3-3215-5259
Copyright 2013 JCN Newswire. All rights reserved. www.japancorp.net