New crypto standard to require protection against power analysis
Second draft of FIPS 140-3 expected to be released soon
- By William Jackson
- Dec 11, 2009
The latest draft of revised federal standards for cryptographic modules, expected to be released for public comment soon, will likely require that encryption tools include protection against differential power analysis (DPA) attacks. Vendors are working now to incorporate such defenses in their products.
The National Institute of Standards and Technology is reviewing the second public draft of Federal Information Processing Standard 140-3. The agency has not set a release date, but it is expected shortly, NIST said.
Although many cryptographic devices, such as smart cards that contain and process encrypted data, already include protection against differential power analysis, the expected requirements in FIPS 140-3 are causing vendors to look at including them in products such as field-programmable gate arrays, which are integrated circuits that can be programmed by users rather than preprogrammed by manufacturers.
“In the last six months, we have seen a lot of attention to this by government and contractors,” said Benjamin Jun, vice president of technology at Cryptography Research, a pioneer in DPA protection that holds patents on a range of anti-DPA technology.
Differential power analysis is a technique for discovering cryptographic keys by monitoring variations in a device’s electrical power consumption and using statistical methods to separate the key from background noise. Although the process of protecting devices is fairly straightforward, the difficult part is defining specifications for the appropriate level of security required on a device.
“A lot of people are looking to NIST to lead this area,” Jun said.
Power analysis was a relatively new technique for cracking codes in single-chip processors when NIST approved the last version of the standard, FIPS 140-2, in 2001. Simple and differential power analysis are mentioned in 140-2 under “other attacks,” but protection against them is not required.
However, DPA now “is one of the bread-and-butter attacks,” said CRI President Paul Kocher.
“A device without countermeasures is basically doing all of its computations in a fishbowl,” Jun said.
Identifying a key with DPA requires some knowledge of the target device and often some reverse engineering. But after an attacker knows what to look for, extracting the key is straightforward, Jun said.
“The statistics for doing this are very forgiving,” he said, “which is very good if you are an attacker, bad if you are a defender.”
An estimated 4 billion cryptographic tools are expected to be shipped with DPA protection this year, and the largest number of them are smart-card chips that protect personal and financial data with encryption. Infineon Technologies, a major supplier of semiconductor chips for the smart-card industry and electronic documents such as U.S. passports, was one of the first chip companies to license DPA countermeasures from CRI in 2008.
But field-programmable gate arrays, which are often used in sensitive communications applications, typically are not protected. The devices are often programmed with an encrypted data stream that could be subject to DPA attacks, and secure computations performed on the chip also could be vulnerable. Because of this, vendors are planning to meet NIST specifications for future products.
FIPS 140 grew out of Federal Standard 1027, General Security Requirements for Equipment, which used the now-outdated Data Encryption Standard algorithm as the standard for encryption. FIPS 140-1 was issued in 1994, with a requirement that it be reviewed every five years. The review and revision process can take several years.
Advances in technology soon outpaced the new version, which requires the Advanced Encryption Standard, and a review of FIPS 140-2 began with a request for comments in January 2005. The original timeline for the new version, which has slipped, called for FIPS 140-3 to be approved by May 2006 and for FIPS 140-2 to be retired in May 2007, although products validated under the previous standard could still be used.
William Jackson is freelance writer and the author of the CyberEye blog.