AFIS Interoperability
Written by Kristi Mayo   

Leaders in the field of latent-print identification are starting
to look for an “enter once, search many” solution

IT IS DIFFICULT to find anyone who denies the power of automated fingerprint identification systems (AFIS) and the benefits they bring to the field of latent-print identification. But ask any expert to list some of the weaknesses of AFIS systems as a whole and one item is likely to appear at the top: interoperability.

The term AFIS interoperability refers to the ability of different AFIS systems to communicate efficiently and effectively with one another. But currently, the proprietary variations between AFIS systems that were created by different manufacturers—and some-times even variations between system versions that were created by the same manufacturer—can cause interoperability problems. In other words, they can prevent examiners from being able to search a latent print encoded on their own AFIS system against another agency’s AFIS system. In a perfect world, experts suggest, a latent-print examiner would be able to capture a latent-print image once, encode it once, and then search it against as many fingerprint or palmprint databases as might be necessary.

Interoperability—which is often rephrased as “enter once, search many”— has been an issue for as long as AFIS systems have been around. But over the last few years, the concept has stepped into the national spotlight.

Currently, the National Academy of Sciences (NAS) Forensic Science Committee is engaged in eight specific tasks related to evaluating the needs of the forensic-science community. The seventh item on their lists of tasks is to examine the interoperability of AFIS systems. In ongoing meetings, the committee has heard from a number of experts on the subject.

Meanwhile, the National Institute of Justice (NIJ) is in the process of forming an AFIS Working Group to carefully study the topic of AFIS inter-operability, as well as a large-scale interoperability study.

Prominent leaders in the field, such as Joe Polski, chief operations officer of the International Association for Identification (IAI), hope these actions may be the beginning of moving the concept of AFIS interoperability into the realm of reality.

“Right now, we are at the problem-identification stage with AFIS interoperability,” said Polski. “There are all kinds of policy issues, technical issues, and legal issues—all sorts of things that would need to be dealt with. But nobody has ever really sat down and looked at this particular issue from a standpoint of What do we have to do to make it work?”

Tenprint vs. latent-print applications

When discussing AFIS interoperability, it is important to make the distinction between the two major AFIS applications: tenprint data and latent-print data. Peter Komarinski, principal of Komarin-ski & Associates, LLC and chair of the IAI’s AFIS committee, explained that the technology of searching tenprints against tenprints with the system has already been advanced to the point of near automation.

“When folks are fingerprinted for tenprint applications, images of all ten fingers are captured,” said Komarinski. “In fact, frequently they are captured twice, because they will take rolled impressions of the individual fingers, then they will take the plain impressions. A tenprint will generally go to the state, and then in many cases it will be forwarded on to the FBI. You have a lot of information to work with, so you have a lot of opportunities for not only making identifications, but also you can link into the FBI to find out if there’s a want or warrant in another state.”

The latent-print side of searching AFIS systems, however, is not so easy or straightforward.

“When you get to the latent-print side of the business, you are not dealing with all ten images. In fact, in many cases, you are not dealing with a very good image. So your opportunity for making that identification on a latent print from a crime scene is pretty much limited to three things: the skill of the examiner who is searching your AFIS system, the sophistication of the AFIS system, and the size of the database on the AFIS system.”

Launching latent-print searches across state lines

In most cases, a latent-print examiner will search their local AFIS system (if available) and then move on to search their state’s AFIS system. The old saying that “local criminals commit local crimes” holds true most of the time. However, it is understood that criminals do not limit themselves to political or geographical boundaries. This fact is especially important to recognize for jurisdictions that lie close to state lines. In such cases, it is highly desirable to have the ability to search the databases of neighboring jurisdictions. Currently, however, getting access to another agency’s fingerprint database is not as simple as it might seem.

As part of a grant from the NIJ, the IAI recently investigated a real-world interoperability situation: Getting the New York City Police Department’s AFIS system to be able to search the New Jersey State Police’s AFIS system.

“This would appear, on the surface, to be a pretty easy process,” said Komarinski. “They are next to each other, separated only by the Hudson River. Both systems have the same vendor, so there shouldn’t be a major problem with conversions and such. But, nonetheless, there are a number of issues that we have run into.

“One issue is that although they use the same vendor, they have different levels of software. Think of a 2008 automobile trying to use equipment from a 2004 automobile. They are pretty close, but not identical. There are also different ways in which the searches take place—different screens that the examiners use.

“That’s on an operational end,” Komarinski continued. “On the legal end, of course, you need agreements between these two entities that they are willing to share their databases and their information, and also that they are willing to protect themselves from any liability that may come up in the future.”

With the difficult issues that were revealed through this project, it may seem that a better answer to AFIS inter-operability would be a larger database containing fingerprint images from people across the country. That database exists, of course: the FBI’s Integrated Automated Fingerprint Identification System (IAFIS)—the federal database that contains the print records of more than 50 million individuals in the Criminal Master File.

Since it was first introduced in 1999, IAFIS has become an increasingly user-friendly method of searching for matches to latent prints. The FBI’s Criminal Justice Information Service (CJIS) Division has worked to make access more widely available to local agencies, and distributes the free Universal Latent Workstation (ULW). This software allows users to remotely encode latent prints and search them against IAFIS.

The system is getting faster, too. Where it used to take hours to get a response after launching a latent search in IAFIS, examiners now frequently get a response in about ten minutes. The FBI’s Next Generation Identification (NGI) system, currently under development, promises to speed up responses even more, and will also add palmprint search capabilities.

In spite of those efforts by the FBI’s CJIS Division, Komarinski said that searching IAFIS “has not been widely embraced by the latent-print community as an easy way of searching latent prints.” Often, he said, after an examiner has taken the steps to search their local or state system—creating a new case number, capturing a latent-print image, making any necessary digital enhancements, and identifying the minutiae points—it becomes necessary to begin the entire process over again in order to get the latent print ready for a search in IAFIS.

“If they have to start the case over again, it becomes very demanding on resource allocation,” said Komarinski. “If your choice is to do more searches on your local system or spend that same time searching the FBI’s system, frequently the decision is not to search IAFIS. There are only so many hours that the examiner has.”

Easy access to IAFIS continues to be an issue for many local agencies, as well, according to Polski of the IAI.

“I have to give the FBI credit. They are working to make access to IAFIS easier and more accessible to the local agencies,” he said. “But not every local agency has access to IAFIS. That is due to a host of problems. For example, we hear from local agencies that want to access IAFIS, but their state will not let them because policies exist that the states must approve local access into IAFIS. So even though the local agency wants to, the state won’t let them.”

Training, resources, and time may also prevent agencies from linking up to IAFIS. “Being able to access the FBI’s database (for searching latent prints) and actually doing it are two different things,” said Komarinski. “They will provide software so that you can search their database, but that does not necessarily mean most agencies will go through the requirements in terms of setting up a separate Originating Agency Identifier (ORI) number, getting a dedicated computer, and then training their people to use it.”

Even if every local agency had a foolproof way of accessing IAFIS, Polski explained that they would not be tapping into a complete database of every fingerprint or palmprint available in the country. The IAFIS database is only as strong as the agencies that contribute to it, and not all information is uploaded to IAFIS. “There is just an awful lot more information sitting in those local AFIS systems than there is at the FBI,” said Polski. “For many years, the FBI would only collect prints from those convicted of more serious crimes. Now, of course, that has changed. Today they will accept anything into their data repository. But many agencies do not send them everything, either, because it is more work to do that.”

In today's world, there is a limited amount of interoperability between independent AFIS systems. A local system (the Local "W", for example, in this diagram) can run a search for fingerprint matches in the AFIS database maintained by its own state (State "A")—and the search can then be directed to the FBI's national IAFIS database. But Local "W" cannot search the AFIS database of another state (State "B"), even if the two states are geographic neighbors. In addition, Local "W" is also probably unable to search other local AFIS systems within its own state boundaries (such as Local "X" in the diagram), not to mention those in another state (Local "Y" and Local "Z"). This limited amount of interoperability severely limits law enforcement's chances of tracking down suspects.

Hurdles standing in the way
of AFIS interoperability

No one, according to both Polski and Komarinski, has gone so far as to draw a detailed picture of what an AFIS-interoperable world might look like. But discussing the concept of AFIS interoperability with experts in the field inevitably elicits detailed descrip-tions of problems that stand in the way of implementing the concept.

A number of these problems were succinctly outlined in September 2007, when the IAI presented a paper to the NAS Forensic Science Committee describing its positions and recommendations on each of the committee’s eight tasks. In that paper, the IAI stated its support for “...the need to pursue the opportunities to improve and increase impression-evidence services via the use of automation technology and electronic networking”. The paper then went on to describe some of the “major stumbling blocks” to AFIS interoperability:

  • Political will—The IAI position paper explains that administrators and managers may be reluctant to move forward with change because of the “...unknown cost/benefits, increased personnel resources, and unknown impact on current AFIS systems.” Because there is no clear picture that shows what kind of system would enable full AFIS interoperability, there currently exist more questions than answers. And that makes it difficult to move forward. Polski posed a few of those questions.

    “If you could launch a search on another agency’s system, could they handle it?” he asked. “Is their system size big enough to just be able to take on all comers? Or if not, how would you throttle the incoming requests? What happens if there is a hit? Is there going to need to be confirmation on the part of the agency that has the print on file? Are they going to need to stop and pull a hard copy, for example?

    “These are the questions that agencies are going to be asking right away.”
  • Connectivity/networking—Here is one obvious question when you start thinking about one agency being able to search another agency’s AFIS system: How do you connect the two systems? For that matter, how would you connect multiple agencies into a broader, more versatile network?

    “The number-one priority is to build a secure infrastructure,” said Stephen Dunbar, product manager with NEC Corporation of America, an AFIS manufacturer. He pointed to the FBI’s CJIS-WAN (wide-area network), the networking component that currently enables local and state agencies to connect to IAFIS, as an example of an infrastructure that could be used for interstate AFIS connections.

    Currently, however, the CJIS-WAN is only being utilized by state and local law-enforcement as a means of connecting to the FBI’s database—not as a method to connect to one another. The FBI has stated that the CJIS-WAN can be used to allow one state to search another state’s fingerprint or palmprint repository, but doing so would require every participating agency to have Memorandums of Agreement (MOAs) in place.

    “Here lies the hurdle,” states the IAI paper. “Obtaining MOAs from all parties is not a simple undertaking and generally requires legal considerations. The expectation of these MOAs being achieved from all parties or in any reasonable timeframe (probably years) is very low. A different approach needs to be considered. National legislation with funding is an alternative.”
  • Consolidation—The IAI paper points out that there currently exists no central location containing all fingerprint records. The FBI’s IAFIS is the closest thing to that kind of collection, but “...given the 50 states and the Federal regulations there are as many reasons why all fingerprint records are not centrally located.” This issue, if unresolved, points the way toward an electronic network (such as the CJIS-WAN or some other alternative) that would connect the multitude of AFIS systems that exist.
  • Maintaining accuracy—When you bring together millions of fingerprint and palmprint records, the technology must be capable of providing the latent-print examiner with a useful, accurate list of candidates. The IAI paper states that the “...interoperability of different AFIS technologies can be addressed either from latent-print images or latent-print minutiae templates… Both can be supported and are supported by the IAI.”

    From the perspective of an AFIS manufacturer, this might be the most important issue to address.

    “The biggest issue with interoperability that I see has to do with the minutiae strings,” said Derald Caudle with AFIX Technologies, Inc. “Every company does the extraction differently—which is understandable. And they use different parameters for their searching and matching. So when you talk about searching from one system to another, I think it is important to do as the FBI has done with IAFIS: allow for an image-only submission. Testing has shown you can get better results with image-only submission. So, for the individual vendors, I think it is important that we always maintain that ability—with minutiae and with image-only submissions.”

    Dunbar with NEC agreed. “There are a lot of people who are shooting for a standard minutiae set, or standard encoding,” he said. “We think that is a really bad idea, because that reduces the unique advantages of each vendors’ different coding systems. It lessens the accuracy of everybody.”

    James Jasinski, executive vice president for federal and state systems with Cogent Systems, Inc., expressed doubts for any kind of standardized encoding. “The accuracy is always going to be highest when the encoding and the algorithm are more closely aligned,” said Jasinski. “If you have a generic encoding and a generic algorithm, you are going to get generic results.”
  • Workload management—In the paper to the NAS, the fifth hurdle to interoperability that the IAI lists is the most brief, but it is also simply to the point: How do you balance your own agency’s needs with the high-priority requests of another agency?

    “The most important resource for an agency is its latent examiners,” said Dunbar. “They are an expensive resource, and you don’t want someone from Virginia tying up the latent examiners in Pennsylvania searching their database. So, then, we need to be able to encode the latent at the source, send it to the destination, match it, and then have the candidate list and all the pertinent images needed for verification to come back to the source. That’s the model the FBI has used in their ULW software.”

    The major AFIS manufacturers have embraced the ULW concept and worked with the FBI to make the latest versions of their systems work with the ULW software. “The ULW application has the ability to translate the minutiae encoding into each major vendors’ proprietary format and send it,” said Dunbar. “So that is how our company is embracing interoperability. We figure it is the best way to send latents back and forth without actually impinging on the other locality’s people.”

    Creating and utilizing more auto-mation technology may also be a way of reducing the increased workloads that AFIS interoperability may create, according to Jasinski. “I think clients and users want to be successful in doing a search, and they want to minimize the amount of work that they have to do. There is a talent associated with latent-print examination. But I think, coupled with the ability to do the encoding, there are automation tools that can achieve the same result without discouraging innovations and inventions in the area,” he said.

    The tools Jasinski suggests would automate the encoding process, similar to the auto-encoding and auto-searching currently available for tenprint searches. “If you look at the history of how tenprints were searched initially in the 1970s, 1980s, and 1990s, there was a lot of manual intervention that was required,” said Jasinski. “Those tools, over time, have been automated so most tenprint searches are done in a completely automated fashion. I think you are going to see the same evolution with the latent world, but you have got to give organizations the incentives to make the investment to do that.”

In a perfect world, there would be a high degree of interoperability between databases of all levels of law enforcement. The Local "W" could search all of the databases in its own state (LOcal "X", for example), as well as those in all of the other states (such as Local "Y" and Local "Z"). Local "W" could also direct searches to its own state database (State "A") and independently access the national IAFIS database. In addition, each of the 50 state AFIS systems would be able to work together without problems or issues. That is how it would be in a perfect world—and such a perfect world may not be too far in the future.

The future of
AFIS interoperability

Those incentives may be difficult to achieve, said Komarinski. He used the analogy of the current financial system—where banks from all over the world are networked, allowing users to access their money on demand. “You can go anyplace in the world and use your debit card, and because there is an interchange fee involved, everybody is happy,” he said. “You are happy because as a consumer you can pay for this. The merchant is happy. The financial institutions are happy. Everybody’s happy. But we don’t have that kind of profit incentive in government. Instead, we have to look for better service and ways of delivering better service.”

Keeping the issue of AFIS interoperability in the national spotlight might gain the concept some momentum.

“I hope that with the work of the National Academies and the position paper the IAI has presented, we are going to see some movement in this now that it is being brought up to a level of national attention,” said Komarinski. “So hopefully it will become more than just a couple of individuals who are saying, ‘This is an issue that really needs to be brought forward.’ Hopefully that will create some real political will to make these kinds of things happen.”

As AFIS interoperability still looms in the realm of conceptualization, Polski said there is still work to be done to determine how beneficial it might be to enable law-enforcement agencies to search the untapped records that lie hidden in other agencies’ fingerprint and palmprint repositories. Some of those benefits may be drawn out by the AFIS Working Group and the interoperability project that has been proposed by the NIJ.

“I think what the NIJ wants to see is, ‘Okay, if in fact we do pursue this, is there a benefit? Are there going to be more criminals identified?’ And that is a good question,” said Polski. “We think there will be more criminals identified. But in the grand scheme of things, when you hook these systems together, maybe there won’t be more identifications.

“I don’t know it for a fact, but I sure think that the accessibility to more data would result in more identifications of criminals.”

INTERNET LINKS FOR MORE INFORMATION:

You can download the IAI’s position and recommendations to the National Academy of Sciences: theiai.org/current_affairs/

And you can get more information and updates about the National Academy of Sciences’ Forensic Science Committee: www8.nationalacademies.org/cp/projectview.aspx?key=48741

 


ORIGINALLY PUBLISHED:
"AFIS Interoperability", written by Kristi Mayo
January-February 2008 (Volume 6, Number 1)
Evidence Technology Magazine
Buy Back Issue

 
< Prev   Next >






Digital-Image Management at Mass Gravesites

SKELETONIZED REMAINS that were carefully unearthed from the desert sands of Iraq tell their own story: the bones of an adult, still dressed in a woman’s apparel, lie supine. The skull is perforated by a bullet hole. Tucked in the space between the ribs and the left humerus is a much smaller skeleton, bones in the skull un-fused, and the fully clothed body partially swaddled in a blanket.

Read more...