Biometric PAD Toolbox Overview [Toolbox]

The TOE may be vulnerable to presentation attacks where attackers attempt to subvert the biometric enrolment or verification by presenting the Presentation Attack Instruments (PAIs). There is a wide range of PAIs that can be used, including natural biometric characteristics, such as dead eyes, or artefacts created from copied or faked characteristics. Using natural biometric characteristics is out of scope of [BIOPP-Module] evaluation and the evaluator shall only use created artefacts to evaluate the TOE.

The toolbox defines the common artefacts for each biometric modality based on publicly available information (e.g. research papers), experiences and knowledge shared among the BIO-iTC members. The evaluator needs to read the [BIOSD] Section 7 as it explains how the evaluator shall use the toolbox during the ATE_IND.1 (Independent testing) and AVA_VAN.1 (Penetration testing) evaluation for Presentation Attack Detection (PAD) in detail. In this evaluation, PAD is being measured using Imposter Attack Presentation Attack Rate (IAPAR), which is a full system review as opposed to a component level review which would evaluate the matching subsystem and PAD subsystem separately.

This overview is originally developed for evaluation activities for FIA_MBV_EXT.3, however, the evaluator can apply the same principles to evaluation activities for FIA_MBE_EXT.3.

Currently toolbox is developed for the following biometric modalities:

The toolbox is intended to be a state of the art set of test and is continually updated. The latest versions of the toolbox for each supported modality must be used by the evaluator at the time the testing begins.

This toolbox overview describes common instructions and components for all toolboxes, rather than repeating the same information in each toolbox. The evaluator shall refer both the toolbox overview and the relevant toolbox to perform independent and penetration testing. If there is conflict between the toolbox overview and a toolbox, the toolbox takes precedence over the overview.

Each toolbox shares the same structure and includes following sections.

To provide flexibility in support for testing the various biometric modalities for PAD, the versioning of the Toolboxes are maintained independently from the versioning of the primary documents ([BIOPP-Module] and [BIOSD]). Each Toolbox is maintained separately within its own GitHub repository so updates targeted to specific modalities can be updated independently as needed over time.

There are several ways to find the Toolboxes that are available.

The simplest method is to go to the GitHub Public Release Packages table from the Biometrics Security iTC homepage. Each available Toolbox will be listed along with a direct link in GitHub to the most recent toolbox package for that modality.

The second method is to go to the Biometrics Security iTC organization in GitHub and find the repositories for each modality there. Each modality has a repository titled <Modality>-Toolbox (where <Modality> would be replaced by a supported modality type). From the home page of the repository, on the right side there is a section titled "Releases". Here you will find all the released versions of the particular toolbox.

Biometric presentation attacks are not all rated the same in terms of their attack potential. This reflects the reality that not all biometric systems can resist all types of attacks. To support this while allowing systems that can protect against stronger attacks to highlight additional capabilities, attacks that target higher attack potentials than AVA_VAN.1 are included as part of the toolboxes.

PAD levels are defined based on the tables in [BIOSD] Section 11, and attacks will be grouped based on the modality and PAD level. PAD Levels may be added to any modality

Higher PAD Levels are inclusive of lower PAD Levels, so a set of tests at PAD Level 2 would incorporate all PAD Level 1 tests. This allows for a reduction in the duplication of attack documentation as test procedures change over time.

As the [BIOPP-Module] relies on the AVA_VAN level specified in the Base-PP, it is possible that the AVA_VAN level for the PAD test may not match the Base-PP. In this case it is up to the evaluator and scheme to determine how to handle the different level claims. The BIO-iTC has determined that there is value in providing PAD tests at higher AVA_VAN levels regardless of the claims by the device that may incorporate the biometric system.

However, the BIO-iTC will explore the possibility of developing a PAD evaluation method at any PAD level that can be mutually recognized under the CCRA, for example, using a multi-assurance evaluation concept that enables the evaluation to conform to a multi-assurance security target.

As explained in [BIOPP-Module], the TOE is the whole biometric system, including Comparison, Decision and Presentation Attack Detection Subsystems. This means in order to successfully overcome the TOE by the use of artefacts, a genuine person (test subject) has to be enroled into the TOE, artefacts have to be created referring the toolbox for the corresponding biometric modality and artefacts have to produce an attack presentation match (i.e. a successful presentation attack).

For all types of testing, there are some common steps/procedures to be followed. These are detailed here.

For independent testing, this guidance is common for all toolboxes. More specific guidance for a specific biometric modality is provided in each toolbox.

This is in addition to guidance in Chapter 7, Common guidance for Independent & Vulnerability Testing.

The evaluator moves to penetration testing only if the TOE passes independent testing. As described in [BIOSD] Section 7, the evaluator shall select those artefacts that show a higher IAPAR during independent testing or higher quality artefacts.

This is in addition to guidance in Chapter 7, Common guidance for Independent & Vulnerability Testing.