Scientists at the National Institute of Standards and Technology are known for sweating the nitty-gritty details of dull but vital standards for everything from bulletproof vests to medical measurement devices and IT encryption.
And now, agency experts are bringing the same precision to computer forensics, fully aware that the final say lies in the notoriously low-tech realm of the judicial system.
NISTs efforts are two-pronged: a library of software applications and programs for testing the reliability of forensics tools.
Since 2001, it has maintained the National Software Reference Library, with support from the Justice Departments National Institute of Justice and law-enforcement agencies. NSRL is an actual library of CDs for 7,120 software applications.
Weve got one of everything we could get our hands on, said Douglas White, a computer scientist in NISTs Information Technology Laboratory.
NIST runs algorithms against the disks to generate digital fingerprints, or hashes, of files, for which it then creates a metadata index. The two together form the Computer Forensic Reference Data Set (RDS) for digital evidence. In a trial, if a court questions the RDS, NIST can prove its authenticity by regenerating the hashes.
Evidence locker
CDs usually arent loaned out. Keeping the originals is very important for evidentiary purposes, said John Tebbutt, another NSRL computer scientist. We do not lend it out because we have to keep it under evidence locker conditions.
As of March, NIST reported, the RDS contained nearly 11 million hashes for three times as many files, and the Web site, www.nsrl.nist.gov, gets nearly a quarter-million hits a month.
Law enforcement agencies and other computer forensics specialists pay $90 a year for quarterly updates.
The RDS speeds investigations by identifying files that can be ignoredsay, a Microsoft Office executable file. It also can highlight hidden and altered files.
NIST staff members say NSRL is such a normal part of the daily work of forensics labs that they dont typically hear success stories from subscribers. Ubiquity can be a better measure: White said the FBI sends copies to all its field offices.
While the basic concept behind it hasnt changed, NSRL is advancing in other ways. NIST is developing a process for hashing network files to address the increasing volume of evidence stored on servers, which are difficult to take into physical custody.
You have to do whats called a live acquisitionacquiring the forensic information from a machine thats actually running, Tebbutt said.
NIST also is working on hashes that operate on storage blocks, a potentially faster, more precise method than the black-and-white file hashes, which can be thrown off by minor file changes. White said block hashes will make it easier to exclude unimportant files. NIST has provided hashes to several states that hope to prevent software tampering in voting machines.