Behind the Scenes of NEC’s Biometrics

The term “biometrics” has been getting a fair amount of airplay these days as news about surveillance and data collection highlight existing privacy concerns. However, the argument in favor of biometrics, which involves using human features to make assumptions about personal identity, is that it can encourage public safety.

In the midst of debates about the ethics of biometrics, however, there is one thing we can all agree on. The use of real-time biometrics platforms to handle complex requests for mission-critical decision-making represents a real “big data” challenge.

IT giant NEC was one of the first large companies to invade the biometrics space and currently claims over 200 deployments worldwide. The company offers everything from fingerprint identification to custom biometrics solutions for military and government use.

As a company familiar with both the hardware and software sides of big data, creating biometrics algorithms and the tools to manage the data they create made NEC the choice vendor for a number of global public service and law enforcement agencies.

A recent use case of one of the company’s biometrics solutions in action raised questions about how the data for real-time biometrics is handled and delivered for nearly instantaneous decision making. The company recently worked with the Texas Department of Public Safety to roll out a real-time identification system that would host biometric details for over 9 million subjects, including fingerprint data, demographics, photos and other materials—all within a mobile platform.

“Identity of suspects lacking proper identification or drivers during suspicious traffic stops can now be confirmed with the NEC Mobile ID at the scene and without the need to remove the officer from his or her patrol,” said Raffie Beroukhim, Vice President of Biometric Solutions for NEC Corporation of America.

We asked John Wise, who handles communications for NEC America,  to describe the how the data is collected, stored and managed for real-time access. He told us that each time a person is arrested for a qualifying crime in Texas, the law enforcement agency is required to fingerprint the individual and send that record to the Texas Department of Public Safety.

According to Wise, this is generally done electronically but can also be done by creating an inked fingerprint card.  Texas DPS either identifies the individual as an already known person or creates a new person record using the existing Criminal Justice Information Systems.  Once identified, the fingerprint record is sent to the Mobile ID system.

Wise told us, that “Mobile ID creates a unique search template from each finger and loads them into an Oracle database on the Mobile ID system.  The database is distributed across several search engine servers for a parallel search process.  It resides in the memory of the servers to speed access to the data.  The number of search engine servers, the processor speed, and amount of memory determine the response time for a search.  As the database grows, more or more powerful servers can be added to the system to maintain a sub 30 second response to identification requests.  The system uses a proprietary fingerprint search engine and fingerprint template generator.”

Wise says that when a local police officer in the field collects a fingerprint from an individual and initiates an identification request, the fingerprint image is sent to the Mobile ID (MID) system.  MID then creates a  search template and compares it to the templates of the known persons on the system and assigns a score to each comparison.  The resulting search score is compared to a preset threshold and if the score is above the threshold, it is deemed an identification.

The proprietary search algorithm and template generation algorithm work in unison to create an extremely accurate system.  This has been tested for accuracy by third party organizations, such as NIST, to independently verify the accuracy of the system.  The officer can receive a red response meaning the individual is definitely identified, a yellow response meaning the identification is probable but needs further investigation or a green response meaning the individual is not in the database.

Wise told us the company plans on expanding the use of biometrics solutions to extend to new markets, including education. Wise says that they’re currently working on a solution that will confirm student identities to prevent cheating.

Related Stories

BI is Becoming Crime’s Biggest Enemy

Big Data Cloud Delivers Military Intelligence to U.S. Army in Afghanistan

Using AI to Combat Terrorism