The FBI expects to have a greatly improved capability for matching fingerprints by late 2010 as part of its Next Generation Identification (NGI) system under the current program baseline and then make improvements and add capabilities on an annual basis, according to the agency’s program manager for NGI.
Whether that improved fingerprint matching is fully implemented by late 2010 or just an initial operating capacity will depend on the agency’s confidence in the solution, Scott Lamoreux tells TR2. Lamoreux says his project baseline agreement with FBI headquarters has some “leeway” built into it, and the target capability could even slip into early 2011.
That flexibility in the schedule is necessary because of the challenge Lamoreux foresees in migrating from the current fingerprint matching capability to the new one.
“One of the interesting statistics is that on any given weekday we do 600 billion individual fingerprint compares on our system,” he says. “So you can imagine with a 24 by seven system in flight, it’s kind of like trying to replace the fan belt on a car going 100 miles per hour without the passengers knowing about it. So the transition is going to be a major undertaking. We have to be very careful with it.”
Early last year the FBI selected Lockheed Martin [LMT] to develop and integrate NGI, which will replace the current fingerprint matching database called IAFIS for Integrated Automated Fingerprint Identification System. The potential 10-year, $1 billion award was initially protested by IBM [IBM], who later agreed to join the Lockheed Martin team and drop its protest. Lockheed Martin also developed IAFIS in the late 1990s.
While IAFIS is a fingerprint matching database, NGI will be a more integrated platform and eventually include other biometric modalities.
Development of NGI follows the need for a more integrated system and increased demand for fingerprint search transactions on IAFIS, which initially were growing 11 percent annually and spiked to 36 percent last year, Lamoreux says. That growth in turn has been driven by fingerprint checks being done by both the Departments of Homeland Security and State related to their respective border security programs, he adds.
Lamoreux says the primary drivers for developing NGI are capacity, scalability and flexibility. The various support systems that are part of IAFIS, such as the work flow processing, user fee billing, and related applications, are independent and have been tied together in a sort of “band aid” approach over the years.
NGI will be an integrated platform with open standards giving it the flexibility to adapt more quickly to new requirements, whether those changes come from Congress or stakeholders, Lamoreux said.
The various components or support systems that make up IAFIS that will now be combined in a more integrated system include:
- AFIS, which is the core fingerprint repository and search capability;
- The Electronic Fingerprint Conversion system that accepts incoming electronic fingerprint transactions and then returns responses
- The Interstate Identification Index, which provides a national repository of biographic and criminal history information for subjects associated with fingerprint records stored within IAFIS;
- The Identification Tasking and Networking (ITN) segment that provides a UNIX client/server-based fingerprint and latent workflow processing system. The ITN maintains a growing file of criminal and civil fingerprint cards that represents over 63 million and 26 million subjects respectively. It also has a national repository, amounting to over 270,000 unsolved latent fingerprints, collected from crime and terrorist scenes; and
- The IAFIS Data Warehouse provides accurate user fee billing data, user interfaces for maintaining user fee administrative data, user fee data auditing, improved queries, and online access for up to two years of historical user fee billing and transaction history data.
Multi-phase Development
The FBI and Lockheed Martin have agreed to a multi-phase six-year development effort for NGI. The program is currently in the design phase. Increment 1 will result in at least the initial capability for the advanced fingerprint matching. In addition to improved accuracy, the FBI expects response times to be improved although Lamoreux says this aspect is a bit of an “unknown” until the various support systems are replaced. He added that response times will improve throughout the implementation of NGI.
As part of Increment 1, Lockheed Martin is conducting a trade study to recommend a vendor for the new fingerprint matching system. Sagem, a unit of France’s SAFRAN Group, is the current fingerprint technology vendor, but is facing competition. Stanford Group emerging technologies analyst Jeremy Grant believes the FBI will stick with Sagem and will also choose technology from at least one more vendor, likely Cogent Systems [COGT] or Japan’s NEC. L-1 Identity Solutions [ID] is also in the hunt.
The “fly-off” is in its final stages and Lamoreux expects a report from Lockheed Martin in March. A final decision on a fingerprint vendor is likely in May or June, he says.
In Increment 2, which will begin while Increment 1 is underway, Lockheed Martin will begin implementing the new integrated business systems that will surround the fingerprint capability. Also in this phase new databases for “individuals of special concern” such as Known and Suspected Terrorists will be created that provide local law enforcement officials with an automated and rapid 10 second response from their wireless fingerprinting devices while in a squad car or other remote locations, Lamoreux says. The rapid response concept is being pilot tested in Minnesota and Ohio. He also expects border security personnel to have a need for the rapid response databases.
The size of this new “risk repository” would be “somewhere under five million records” made up of the “worst of the worst types of people,” Lamoreux says.
The fully integrated business system implementation is slated to be completed in Increment 3, which wraps up in the 2012 timeframe, Lamoreux says. This period is also when a new biometric modality, the ability to identify latent palm prints, will be added to the system, he adds.
Increment 4 will introduce facial recognition technology as well as the ability to identity individuals and the groups they belong to based on scars, marks ant tattoos, Lamoreux says.
“Scars, marks and tattoos identify a lot about why a person is here, who’s with them, where they are from,” he says. They “are expected to make a big difference to local law enforcement.”
If the schedule holds, Increment 4 would finish in 2013. Increment 5 is when iris recognition technology would be integrated into NGI as well as initiating an effort to fuse the biometric technologies to get as close as possible to a 100 percent identification solution, Lamoreux says.
Increment 5 would end in 2014 and be followed by the final increment, which is the validation of performance phase and a wrapping up of the biometric fusion effort.
The entire program has the flexibility to make adjustments, such as bringing iris recognition capabilities forward, based on demand, Lamoreux says.
Each successive increment overlaps the previous one in part to keep to an aggressive schedule but also to keep the subject matter expertise fully engaged throughout the development and implementation cycles of the program, Lamoreux said.
For each biometric modality that is introduced trade studies will be completed. However, Lamoreux says that while there has been some pre-costing of these, final costs for procuring those solutions haven’t been determined.
The FBI’s currently published accuracy rate of IAFIS is 95 percent, which means that there is a 95 percent probability of finding a person in the database from an inbound search request regardless of the quality of the fingerprint. Combining that fingerprint search with other types of identification such as name check further improves those odds, Lamoreux says. Under NGI the goal is a 99 accuracy rate, he says.