Time for Multi-Modal Biometrics at Border Security Checkpoints
Author(s): John Kendall, Posted on April 15th, 2016
Border security today is facing a perfect storm of challenges that requires every tool available. But when it comes to leveraging biometrics, border security agencies often cling to outdated technologies and inaccurate assumptions.
Border Security Challenges
Recent terrorist attacks in Paris and Brussels put pressure on already stressed border security agencies. And the problem isn’t limited to terrorism. According to the UN Office on Drugs and Crime, people trafficking is the world’s fastest growing global crime. According to the UN World Tourism Organization (UNWTO), international tourist arrivals increased by 4% in 2015 to reach a record 1.2 billion. And with border agency budgets flat in most countries, many border agencies are handling greater workloads with the same number of people and facilities.
Despite the increased security and higher volumes, travellers expect faster and easier border clearance. Furthermore, stolen passports, fraudulent travel documents, resource constraints and information overload add to border security headache and create an untenable situation for both travellers and border agencies.
Biometrics can help
New ePassports include facial biometric data on the chip, so biometrics can automatically detect stolen or forged passports by authenticating the traveller against the rightful holder of the travel document. Border agencies can also use biometrics to check the traveller against a watch list of known “most wanted” persons to identify persons of interest when entering or leaving the country. And automated clearance eGates are capable of performing these checks quickly and accurately.
Border security solutions employing biometric technology are used in many countries today including the US, UK and Australia. But these biometric solutions are little different from those deployed 15 years ago and continue to exhibit the same shortcomings. In particular, most of the current biometric solutions are unable to detect individuals travelling under multiple identities and travel documents. This is a vulnerability that can be exploited by terrorists and other criminals to avoid detection when travelling internationally. If an individual is able to obtain a new passport (perhaps from a different country) under a new “clean” identity, then the chances of getting stopped by border security officers are very small.
The problem is that the types of biometric captured at most border crossings aren’t well suited for near-real time searching against very large databases (e.g., biometric records of all travellers who previously entered or exited the country).
The International Civil Aviation Organization (ICAO) Document 9303 defines international standards for machine readable travel documents, like ePassports. The standard provides for the storage of three different types of biometrics on the chip – face, fingerprint and iris. Facial biometrics are mandatory, but fingerprint and iris modalities are options.
Facial biometrics work very well for performing a 1-to-1 verification of the traveller to the facial image stored on the chip – it is quick and accurate. But facial biometrics are not as well suited for performing 1-to-many searches against a large database of biometric records because of the large number of false matches and false non-matches. For example, if a traveller’s face is compared against the faces of 100 million previous travellers, the facial matching system is likely to return a long list of possible matches against records with similar faces. A border agent then needs to manually review the possible matches to eliminate all the false matches. Not a problem if you have time, but when facing a queue of tired and frustrated travellers time is an unobtainable luxury.
Because of the relatively low accuracy of facial biometrics, a number of countries have elected to collect and match fingerprints at the border crossing. Fingerprint image analysis detects far more feature points (or minutiae) in a single fingerprint than facial biometrics detect in a face. And fingerprint biometric matching performs a far more mathematically complex comparison of those feature points (e.g., location, ridge direction, and distance to neighbouring feature points). As a result, fingerprint biometrics are far more accurate than facial matching. In fact, it is possible to perform 1-to-many searches against a large database of fingerprint biometric records with very few false matches and false non-matches.
Sounds good, but there is a catch.
In a border crossing situation, the biometric matching needs to be completed in at most a couple of seconds – or near-real time. Since fingerprint matching is computationally intensive, near-real time, large-scale fingerprint matching requires significant processing resources – which can be very expensive. So fingerprints work well for 1-to-1 authentication and 1-to-few watch list checks, but fingerprint biometrics are too costly to perform near-real time searches against massive databases (such as the biometric records of all previous travellers). And without that capability, a known suspect travelling under a new identity and travel document can slip through the border undetected.
Iris – best of both worlds
Iris biometrics offers the advantage of very fast and efficient matching with an accuracy similar to fingerprints. As a result, it is possible and cost effective to perform near-real time iris biometric matching against very large iris databases. So how might iris biometrics be used in the border security environment?
When a traveller enters or exits the country, the border agency captures an image of the iris. This is a simple process that takes a high resolution picture of the eye from up to 2 meters away – much like taking a photo of the face. Once the iris image is captured, the unique patterns of the iris can be quantified and searched against the entire database of previous travellers to determine whether or not that iris has been seen previously.
Iris biometrics represents the best defence against individuals who attempt to enter a country using multiple identities and will go a long way towards tightening border security without delaying the border clearance process. Iris biometrics are not as well known or understood by the public as facial or fingerprint biometrics, but iris biometrics are used for border clearance in the UAE and iris is the favoured modality for large-scale civil applications – like national identity. For example, iris is the primary biometric used for the 110 million person Mexico National ID as well as the 1 billion person India National ID.
Most border agencies try to weather the perfect storm of border security challenges using traditional biometric technologies that only address part of the security risk. With heightened security threats, there is a pressing need to expand border crossing solutions to leverage the power and cost efficiency of iris biometrics. Face and fingerprint biometrics still have a place, with many existing face and fingerprint biometric watch lists, but the time for multimodal biometrics (using face, fingerprint and iris) has arrived.