Biometrics Field Trial Evaluation Report

Section 10. Lessons Learned from the Field Trial

10.1 Strengthening identity management of clients

Biometrics can help strengthen identity management of clients at CIC, because it can reliably link each client to an immigration decision. Implementing biometrics would strengthen the integrity of immigration programs and increase CIC’s confidence in clients’ identity.

Biometrics can assist greatly in visa program integrity. Out of 14,854 visa applicants, 394 biometric highly accurate matches were found using an automated system. Even under the limited conditions of the field trial, biometrics helped CIC discover that one visa client had changed their identity after arriving in Canada and had claimed refugee protection. In a situation where decision makers would have real‑time access to matching results, biometric matching could assist in validating identity and reducing the incidence of identity fraud. The same automated matching capability could be used to check previous immigration enforcement activities and criminal records.

Using biometrics can help with client identity management across different CIC programs. During the field trial, 12 visa applicants came to Canada and made refugee protection claims. While 11 of them used the same name as on their visa, one did not. This case is considered fraud.

Having an electronic photo of the visa holder has value for verifying identity at the port of entry.
The combination of both biometrics (photos and fingerprints) proved to be better than either used alone. Facial recognition alone failed to identify two sets of photos above the recommended threshold. Fingerprint recognition alone failed to find two matches above recommended threshold. Because the two facial recognition false non‑matches were not for the same people as the false non‑matches for fingerprints, when combined together, using both biometrics yielded all matches.
While fingerprints are, as expected from the large body of research, a much more accurate biometric, they alone cannot assist in all cases. During the field trial, 918 clients arrived at a port of entry and presented a single fingerprint that was compared against their previously enrolled 10‑fingerprint set. Forensic experts reviewed each system‑generated match, along with the images of all fingerprints and photos enrolled. They judged 36 fingerprint records (3.9%) to be of too poor quality to accurately assess whether or not there was indeed a match. In a fully deployed system, the ability of an officer to compare the retrieved photo with the live traveller will help in these cases.
Verifying biometrics at the border may have deterred the entry of some clients at ports of entry participating in the field trial. While the number of field trial clients actually exceeded expectations, analysis shows that the portion detected at the participating ports of entry was much lower than expected (only 10% of field trial clients were detected at the participating ports of entry). Even allowing for mitigating factors such as the recent direct flights from Hong Kong to Toronto and equipment issues at the airport, this is an extremely low number.

10.2 Client service

Service levels: Current service levels at visa offices were maintained while collecting biometrics, but only through the use of additional human resources. The time to enrol and verify were, on average, judged to be acceptable and did not impede the flow of business. Nonetheless, some existing facilities had to be renovated in order to accommodate biometric enrolment equipment such as fingerprint readers. Mandatory in-person enrolment of biometrics at the visa offices could require significant investment in facilities (such as enlarged waiting rooms, more service counters) and human resources to accommodate the increased traffic.

Photos: The field trial showed that it is definitely feasible to obtain photos good enough for facial recognition from most clients, at least in urban centres like Hong Kong and Seattle, if CIC’s new visa photo standards are enforced consistently over many months. However, this has proven to be stressful for both employees and clients.

Demanding that photos comply with CIC photo standards requires an extensive information campaign in every locale to inform potential clients and photographers of the changes months before any full implementation. Posting on the Internet and creating brochures, as was done during the field trial, would need to be augmented by other media and means of communication, and by providing posters and brochures to officers and to travel agents, and perhaps even by holding press conferences. At a minimum, consideration should be given to placing the information on photo standards more prominently on CIC’s website. Visa office employees also suggested that showing a video in the waiting room might be useful for walk‑in clients. In some cases, the presence of the photo guidelines poster in the visa office waiting room was enough to have people quickly realize that they had to return with better photos.

Given the additional pressure that requiring compliant photos caused for operations and the extensive communication requirements, consideration may need to be given to having photos taken at the visa office. This way, the client would be certain to have photos that would be acceptable to CIC, and CIC would be certain that the photo has not been tampered with.

Client feedback: Giving fingerprints to a CIC or CBSA official was not perceived as an issue for clients. Once clients were given an explanation for why they were being asked to provide their fingerprints, they seemed satisfied. For any full implementation of biometrics, CIC should make every effort to provide a clear reason for asking for a client’s biometric through clear communications in order to maintain client satisfaction.
Client facilitation: During the field trial, 364 clients were seen more than once by CIC. Processing of these repeat clients could have been facilitated using biometrics—either by not requiring them to enrol their 10 fingerprints at a subsequent encounter or by requiring them to give only one fingerprint to verify their identity instead of resubmitting their biographical and possibly other personal identifying information.
Not a 100% solution. Both facial recognition and fingerprint systems on their own failed to enrol a very small number of clients. Client age, not gender, was a factor in the effectiveness of fingerprint recognition in the operational environment. During the field trial, it took more effort and time to get good‑quality fingerprints from older clients. Field trial analysis showed that the time to take fingerprints is directly correlated to age. Specifically, it took significantly longer to enrol 10‑fingerprint sets of clients over 70 years of age compared with younger clients. Fingerprint enrolment time becomes longer with each subsequent age group. At the port of entry, it was found that it took longer to verify one fingerprint of clients over 60 years old. If fingerprint recognition technology is to be used for full implementation, consideration should be given to having an upper age limit.

10.3 Privacy

The field trial showed that it is possible to collect biometrics and analyse them in a way that protects clients’ privacy. Taking privacy into account early in the operational concept design stage was effective. Full implementation should duplicate as many of the privacy measures taken in the field trial as possible and applicable. Notably, any future biometric implementation would consider the physical security of data repositories, encryption, secure transmission, secure storage, audit, and investigation.

10.4 Forward‑planning lessons

10.4.1 Operations

The right workflow is crucial to the successful implementation of biometrics. While every CIC visa office has a slightly different workflow and needs flexibility, the field trial has shown that all the clients’ biometrics must be collected in the same step and be saved to their record, preferably with the client in attendance. When this does not happen, there is room for administrative error. During the field trial, when the photo scanning and the fingerprinting of the client occurred separately, there were six instances of the client’s fingerprint being saved in another client’s file. This was discovered when the clients’ fingerprints verified at port of entry was found not to match the fingerprints on file.

The field trial also showed that biometric information collection needs to be incorporated at the correct point in the visa office’s existing workflow. If not, biometric enrolment can significantly affect productivity. Since every visa office has a slightly different workflow, CIC NHQ should provide common universal principles but leave flexibility for local workflow management in any full implementation. In addition, operations management and queue management training with considerations for biometrics should be offered to immigration program managers before any full implementation.

The “ghosting” effect experienced during the field trial, as well as the cases of administrative error due to workflow issues, suggests that any full implementation of biometrics should include an initial period where no administrative actions are taken. This period should be used for quality control of both the system and the work processes.
User training. For any full implementation, a formal training program and schedule will be required for each user and to ensure that each user has time away from duties to be trained and is covered at work. Coaching the users after the training and having the coaches stay for the initial period of the system going live is also highly recommended. The field trial showed that users at sites where coaches stayed longer had a better understanding of the system and retained their training better. It is also highly recommended that basic training be given on what constitutes high‑quality photos and fingerprints.
Experience counts: Of the four different user groups who participated in this field trial, the group who made the fewest errors was the group in Hong Kong who was specifically dedicated to biometric enrolments. They enrolled a multitude of clients every day and had an efficient work process where they made the most use of the system wait time.

Experience also makes a difference when it comes to enrolling a client’s 10 fingerprints. The user group with the lowest median time for fingerprinting was the group that enrolled the most clients. There is a direct correlation between the number of clients a user group enrolled and the median enrolment time.

Any full implementation would require human expertise to confirm the system‑suggested matches. While the biometric system is accurate in most cases, suggested matches must still be confirmed. The number of suggested matches for human review can be adjusted depending on operational and program integrity requirements. It was found that while CIC employees could be trained to recognize faces, fingerprint match confirmation required a trained forensic specialist.
Extra human resources will be required at visa offices if clients are enrolled at these offices. During the field trial, the visa offices were able to manage enrolling their walk‑in clients and doing the extra back‑office work with the additional resources assigned from the project. The impact of biometrics was seen to be greater on a smaller visa office.
Using chips in CIC visas adds technical and operational complexity, both at the visa office and at the border. From an operational point of view, chips also add an extra step to the visa issuance process. This step can be particularly time consuming if visa offices receive defective chips. Quality control of the chips also becomes an issue.

The field trial also confirmed that chip collision would be an issue if CIC were to use chips in visas. In the only instance where employees mistakenly put the biometric field trial visa in an e-passport, they were no longer able to read the information on the CIC chip.

Using chips also requires extra equipment (chip readers and PC tablets) to be bought, installed and maintained. There were issues with these peripheral devices at the ports of entry during the field trial. In addition, in workstations with limited areas such as the primary inspection line at the VIA, ergonomics emerged as an important issue. The recommendation for any full implementation is to have an integrated system so that the need for chips, chip readers and PC tablets is eliminated.

10.4.2 IT / Biometric systems

For any full implementation, employees expressed desire for the following features:

Time to save fingerprints to the biometric system should be faster than during the field trial. Employees at every field trial site cited this.
A user fingerprint reading feedback screen that provides feedback from the system to let officers know how they can improve their score, in addition to the other feedback the system provided during the field trial. The field trial fingerprint feedback used most was the numerical score. The colour map was helpful for determining which finger needed to be pressed harder against the fingerprint reader
Option of a simple visual feedback to the client when they are enrolling their 10 fingerprints.
The ability to select the best fingerprint set if multiple sets were captured (retakes). In some cases the quality of the retake was worse than that of the original set.
Performance reporting capabilities should be defined in advance and built into the design, using lessons learned from this field trial. During the field trial, even though IT systems had reporting capabilities, the design did not anticipate how user behaviour would affect some measurements. For example, fingerprint verification start time was triggered by the chip in the visa being detected. If a traveller’s passport was left on the chip reader during examination, the fingerprint reader timed out a number of times and verification time, calculated to end with the fingerprint being captured, may in fact be less than the time shown in the IT reports. A similar situation occurred with 10‑fingerprint enrolment, the starting time being triggered by the user pressing “fingerprint capture.”
Strong local IT support is important to having a successful implementation, especially during the installation and initial launch stage. During the field trial, IT support required at the visa offices was minimal, but the chip readers and tablet PCs at the ports of entry required considerable IT attention from CIC Ottawa and CIC Regional Office to be resolved.
For high‑quality fingerprints, the glass of the fingerprint reader must be cleaned regularly but not necessarily after every client.
For any full implementation at port of entry immigration secondary, there should be one fingerprint reader at each workstation. During the field trial, time was lost when officers and clients had to go to a workstation with a fingerprint reader or when those stations with fingerprint readers were unavailable. Any equipment should be secured to the workstation, and officer safety and ergonomics should be kept in mind.

<< Previous | Contents | Next >>