IARPA Awards SRI International Multi-Year $12.5 Million Contract To Address Vulnerabilities In Current Biometric Security Systems | SRI International

Toggle Menu

IARPA Awards SRI International Multi-Year $12.5 Million Contract To Address Vulnerabilities In Current Biometric Security Systems

MENLO PARK, Calif. June 27, 2017 – SRI International has been awarded a four-year $12.5 million contract by IARPA’s (The Intelligence Advanced Research Projects Activity, National Intelligence Directorate) Odin Program to research and develop “dynamic biometrics” able to better detect attempts to evade or deceive biometric security systems, such as fingerprint, iris and face scanners. The ability to detect these “presentation attacks” addresses a critical weakness in current generation biometric security systems and can significantly expand biometric use cases.

Bhatnagar

The market for biometric sensors is expected to grow to $24.4 billion by 2020, driven by initiatives taken by governments and businesses across the world to adopt fingerprint, iris, and face biometrics systems for a broad range of use cases such as travel checkpoints, facility access points, and identity verification and cyber-authentication.  However, the inability to detect presentation attacks (which attempts to deceive biometric security systems are often called) relegate biometric use cases to: 1) relatively low risk applications such as unlocking one’s phone, or 2) applications in which a human guard is present to spot and stop overtly suspicious activity.

“Biometric systems will not be successful unless technical solutions providing reliable but low cost detection of currently unknown presentation attacks can be developed,” said Jeffrey Lubin, Ph.D., senior research scientist at SRI and Principal Investigator for SRI’s Odin team. “We believe the dynamic biometrics developed in this IARPA Odin Program, which includes deep collaboration between SRI’s computer sciences and biosciences researchers, will greatly expand the range of uses for biometric systems, benefiting people, businesses and governments around the world.”

SRI International will forge and deliver a prototype Multi-physiological Joint Optimization and Liveness Nuances for Identity Ratification system (MJÖLNIR) to defeat known and unknown presentation attacks. To succeed, SRI has assembled a strong multidisciplinary team with depth and diversity of expertise in human physiology (SRI’s expertise in biosciences, signal analytics and machine learning), dynamic 3D imaging (Compact Imaging), 3D shape and behavioral dynamics modeling (CMU, SRI), and system development, evaluation and deployment (SRI).

How Dynamic Biometrics Work:

  • Conventional biometrics rely on static surface images of human tissue. Scanning tissue below the skin surface to detect dynamic physiological information that conventional scanners do not reveal is a significant element of the SRI team’s approach in the Odin program.
  • Under the Odin Program, the SRI team is researching innovative dynamic biometrics systems and techniques to dramatically improve the presentation attack detection capabilities of biometric systems by imaging, measuring and analyzing real-time physiological responses of living tissue to external and internal stimuli. 
  • By analyzing such factors as changes in heart rate, perspiration and blood flow both within tissues being used for biometric identification and across other body regions, the system will reliably detect whether these biometric tissues are real or being faked.

This research is based upon work supported in part by the Office of the Director of National Intelligence (ODNI), Intelligence Advanced Research Projects Activity (IARPA), contract number 2017-17022400004.

The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of ODNI, IARPA, or the U.S. Government. The U.S. Government is authorized to reproduce and distribute reprints for governmental purposes notwithstanding any copyright annotation therein.