Biometric data processing method, associated computer system and program.

The method uses functional encryption with masked polynomial functions to secure biometric data calculations, addressing privacy concerns and enabling efficient, secure biometric authentication without exposing reference data.

FR3148305A1Pending Publication Date: 2024-11-01IDEMIA PUBLIC SECURITY FRANCE
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
FR2023004129
Authority / Receiving Office
FR · FR
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-04-25
Publication Date
2024-11-01

AI Technical Summary

Technical Problem

Existing biometric identification and authentication systems using functional encryption algorithms leak information about reference biometric data during distance calculations, limiting the number of identifications or authentications and compromising privacy.

Method used

A method involving functional encryption of biometric data using a polynomial function of degree 1 or 2, masked by a primary mask, where biometric data is represented as vectors with additional masking elements, and decryption keys are generated on the fly, ensuring secure distance calculations without exposing clear distances.

Benefits of technology

This approach maintains privacy by preventing the calculation of clear distances, thus securing biometric data while allowing multiple authentications without storage of reference data, enhancing security and efficiency.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 00000000_0000_ABST
    Figure 00000000_0000_ABST
Patent Text Reader

Abstract

A method for processing biometric data comprising: - Functional encryption (E104) of a test biometric data set which is a vector comprising n test components representing a biometric trait of a candidate individual, and at least one other non-zero test component representing a first masking element, - Obtaining a score (E106) representing a distance, between the test biometric data set and a reference biometric data set, in a form masked by a primary mask, by functional decryption of the encrypted test biometric data set, using a functional decryption key for a polynomial function of degree 1 or 2 parameterized with said reference biometric data set,The reference biometric data is another vector comprising n reference components representing a biometric trait of a reference individual and at least one other non-zero reference component representing a second masking element. Figure for the abbreviation: Fig. 2.
Need to check novelty before this filing date? Find Prior Art

Description

Title of the invention: Method for processing biometric data, associated computer system and program.

[0001] The invention relates to a method for processing biometric data, for the comparison between a test biometric data and at least one reference biometric data.

[0002] Identification or authentication schemes are already known in which a user, also called a candidate individual, presents to a trusted processing unit, for example to a unit belonging to customs, an airport, etc., a biometric data newly acquired on the user which the unit compares with one or more reference biometric data recorded in a reference database to which it has access.

[0003] This reference database groups together the reference biometric data of authorized individuals, also called reference individuals, such as passengers of a flight before boarding.

[0004] From application FR3123130A1, a biometric data processing method intended to be used to identify or authenticate a candidate individual, and based on a clever mechanism based on a functional cryptographic encryption algorithm.

[0005] The functional encryption cryptographic algorithm may be as described in "Function-Hiding Inner Product Encryption is Practical", Sam Kim, Kevin Lewi, Avradip Mandai, Hart Montgomery, Arnab Roy, and David J. Wu SCN 2018: 544-562.

[0006] For each reference biometric data, a functional decryption private key is generated, so as to constitute a database of functional decryption private keys.

[0007] A biometric proof data representative of a biometric trait of the individual candidate is encrypted using a functional encryption public key.

[0008] To calculate a distance between the test biometric data and a reference biometric data, the encrypted test biometric data is decrypted by functional decryption using the functional decryption private key generated from the reference biometric data.

[0009] Thus, one can be satisfied with a reference database containing private decryption keys, each associated with a reference biometric data (it is not necessary to keep the reference biometric data), but not allowing one to go back to this reference biometric data.

[0010] This is also very advantageous in terms of data storage size.

[0011] Such a solution is satisfactory, but the calculated distances between a data point Reference biometrics and several proof biometric data sets can leak information regarding the reference biometric data.

[0012] To limit the amount of information that could leak, the number of functional decryptions that can be executed using the same functional decryption private key is restricted.

[0013] This restriction limits the number of identifications or authentications that can be performed with a reference database.

[0014] It would therefore be desirable to have a simple, reliable, secure, privacy-respecting solution that remedies this drawback for the identification and / or authentication of an individual.

[0015] To this end, the present invention proposes, according to a first aspect, a method for processing biometric data, the method comprising the implementation by a system of steps of: - Functional encryption of biometric proof data using a functional encryption key, - For at least one reference biometric data point, obtaining a score representing a distance between the test biometric data point and the reference biometric data point, by functional decryption of the encrypted test biometric data point, using a functional decryption key for a polynomial function of degree 1 or 2 parameterized with said reference biometric data point, the method being characterized in that: - The biometric proof data is a vector comprising n proof components representing a biometric trait of a candidate individual, with n a natural number strictly greater than zero, and at least one other non-zero proof component representing a first masking element, - The reference biometric data is another vector comprising n reference components representing a biometric trait of a reference individual and at least one other non-zero reference component representing a second masking element, and - The score obtained by functional decryption represents the distance between the test biometric data and the reference biometric data in a form masked by a primary mask.

[0016] According to advantageous and non-limiting features: - the primary mask has the value of the result of applying to at least one other test component the polynomial function of degree 1 or 2 parameterized with at least one other reference component; - The process further includes determining, by random or pseudo-random sampling, a value of at least one other component from among the at least one other reference component and the at least one other test component; - The process further includes an enrollment step comprising the generation of said functional decryption key for said polynomial function of degree 1 or 2 parameterized with said reference biometric data, from a master key and said reference biometric data; - said functional decryption key for said polynomial function of degree 1 or 2 parameterized with said reference biometric data, is generated by application of said polynomial function of degree 1 or 2 between said master key and said reference biometric data; - the enrollment step includes the generation of said functional encryption key from said master key; - the enrollment step includes the generation of said master key, the obtaining of at least one reference biometric data, and, for each reference biometric data obtained, the generation of the functional decryption key for said polynomial function of degree 1 or 2 parameterized with said reference biometric data, so as to constitute a database of functional decryption keys; - the functional encryption step includes obtaining said biometric proof data from the biometric trait of the individual candidate using biometric acquisition means of the system; - the polynomial function of degree 1 or 2 is a scalar product; - The process further includes, for i being equal to 1 and 2, the implementation of the following step by a device of index i of the system: Generation of a partial result of index i from the score and a demasking data of index i associated with the primary mask; - the respective index devices 1 and 2 are distinct; - the partial results of respective indices 1 and 2 allow the calculation of a control result indicating whether the test biometric data corresponds or not to the reference biometric data; - the control result is equal to the sum of the partial results of respective indices 1 and 2; - the functional encryption of the biometric proof data is implemented by a device of the system, separate from the devices of indices 1 and 2, and / or the control result is calculated from the partial results of respective indices 1 and 2 by an output device of the system, separate from the devices of indices 1 and 2.

[0017] According to a second aspect, the invention provides a computer program comprising code instructions for executing a processing method for biometric data as defined above, when this program is executed by a system.

[0018] According to a third aspect, the invention proposes a computer-readable storage means on which a computer program includes code instructions for the execution of a biometric data processing method as defined above.

[0019] According to a fourth aspect, the invention proposes a biometric data processing system, the system being configured to implement the following steps: - Functional encryption of biometric proof data using a functional encryption key, - For at least one reference biometric data point, obtaining a score representing a distance between the test biometric data point and the reference biometric data point, by functional decryption of the encrypted test biometric data point using a functional decryption key for a polynomial function of degree 1 or 2 parameterized with said reference biometric data point, and the system being characterized in that: - The biometric proof data is a vector comprising n proof components representing a biometric trait of a candidate individual, with n a natural number strictly greater than zero, and at least one other non-zero proof component representing a first masking element, - The reference biometric data is another vector comprising n reference components representing a biometric trait of a reference individual and at least one other non-zero reference component representing a second masking element, and - The score obtained by functional decryption represents the distance between the test biometric data and the reference biometric data in a form masked by a primary mask.

[0020] This system can be configured for the implementation of each of the embodiment possibilities envisaged for the biometric data processing method as defined above.

[0021] Of course, the different features, variants and embodiments of the invention can be combined with each other in various ways insofar as they are not incompatible or mutually exclusive.

[0022] Other features and advantages of the present invention will become apparent from the description given below, with reference to the attached figures which illustrate examples of embodiment without any limiting character.

[0023] In the figures:

[0024] [Fig. 1] schematically represents a preferred embodiment of a system for the implementation of a process according to the invention;

[0025] [Fig.2] illustrates the steps of an embodiment of a process according to the invention.

[0026] With reference to [Fig.1], a biometric data processing system 1 has been schematically represented for the implementation of a biometric data processing method for the authentication / identification of candidate individuals.

[0027] In one embodiment, system 1 implements an authentication of a candidate individual, that is to say compares the so-called test biometric data (freshly acquired on the candidate individual), to a single reference biometric data, supposed to come from the same individual, in order to verify that the individual from whom the two data were obtained is indeed the same.

[0028] In another embodiment, system 1 implements identification of the individual candidate, i.e. compares the test biometric data to all reference biometric data in a database, in order to determine the identity of the individual candidate.

[0029] Typically, the test biometric data and each reference biometric data are vectors in the same vector space. The test biometric data comprises n test components representing a biometric trait of a candidate individual, where n is a natural number strictly greater than zero. Each reference biometric data comprises n reference components representing a biometric trait of a reference individual.

[0030] This system 1 is equipment owned and controlled by an entity with whom authentication / identification must be carried out, for example a government entity, customs authority, company, etc. In the remainder of this description, we will take the example of an airport, with system 1 typically intended to control passenger access to a flight before boarding.

[0031] The system 1 includes a control device 2, at least one pair of participating devices 3, and a trust device 4 and an enrollment device 6.

[0032] The control device 2 includes a processor 20, a communication interface 22 for communicating with the trust device 4 and with each participating device 3, a memory 24 and a biometric sensor 26.

[0033] The processor 20 is configured to implement certain steps of a process that will be described later. The processor can have any structure. The processor comprises one or more cores, each core being configured to execute the code instructions of a program in such a way as to implement the aforementioned steps.

[0034] The communication interface 22 is for example of the wireless radio type, and uses any communication protocol (Wi-Fi, Bluetooth, etc.).

[0035] Communications between the communication interface 22 of the control device 2 and communication interfaces 32 of participating devices 3, may use identical or different protocols for communications between the communication interface 22 of the control device 2 and a communication interface 42 of the trusted device 4.

[0036] The memory 24 is adapted to store data manipulated or produced by the processor 20. The memory 24 is of any type. Conventionally, the memory comprises volatile memory for storing data temporarily, and non-volatile memory for storing data persistently, that is, in a way that retains the data when the non-volatile memory is powered off.

[0037] The biometric sensor 26 is configured to acquire biometric data relating to individuals. For example, the biometric sensor includes a camera configured to acquire images showing the face of a candidate individual and to extract proof components from such images. Alternatively or in addition, the biometric sensor includes a fingerprint sensor and / or an iris scanner. Various image processing techniques for extracting representative components of the individual, typically proof components representative of the candidate individual, are known to those skilled in the art. By way of non-limiting example, the extraction of representative components of the individual may include the extraction of particular points or a facial shape in the case where the image is an image of the individual's face.

[0038] In one embodiment, the control device 2 further comprises a gate 28 that can be closed to prevent an individual from accessing a secure area, and opened to allow such access. The processor 20 is configured in this case to control the opening and closing of the gate 28. For example, the control device 2 is located in an airport, and the secure area is a boarding area; in this particular application, the individuals wishing to access the boarding area are the passengers of a flight, whose identity must be verified before boarding.

[0039] Each participating device 3 comprises a processor 30, a communication interface 32 for communicating with the control device 2, with the trust device 4 and with the enrollment device 6, and a memory 34. The information provided above concerning the processor 20, the communication interface 22, and the memory 24 is also applicable to the processor 30, the communication interface 32, and the memory 34.

[0040] The communications between the communication interface 32 of a participating device 3 and the communication interface 22 of the control device 2 may use protocols identical or different from the communications between interface 32 of the participating device 3 and communication interface 42 of the trust device 4 and / or communications between interface 32 of the participating device 3 and a communication interface 62 of the enrollment device 6.

[0041] The participating devices 3 are distinct from one another. In what follows, we will detail an embodiment in which the participating devices 3 are distinct from the control device 2, the trust device 4, and the enrollment device 6, as shown in [Fig. 1]. However, in other embodiments, it may be envisaged that the control device 2 and / or the enrollment device 6 constitutes one of the participating devices 3.

[0042] Each memory 34 stores a reference database, which is a functional decryption key database. Each functional decryption key in said database is associated with a reference biometric data point relating to a previously enrolled individual, also referred to as the reference individual. The reference biometric data points are not stored in the reference database. A functional decryption key cannot be used to retrieve the associated reference biometric data point. The reference biometric data points are thus protected from confidentiality.

[0043] The trust device 4 has the function of generating a cryptographic key, referred to as the master key, a primary mask, masking elements associated with the primary mask, and demasking data associated with the primary mask, which are used by other devices in the system. The trust device 4 comprises a processor 40, a communication interface 42 for communicating with the control device 2, with the enrollment device 6, and with each participating device 3, and a memory 44. The information provided above concerning the processor 20, the communication interface 22, and the memory 24 also applies to the processor 40, the communication interface 42, and the memory 44.

[0044] Communications between the communication interface 42 of the trust device 4 and the communication interface 22 of the control device 2, communications between the communication interface 42 of the trust device 4 and the communication interface 62 of the enrollment device 6, and communications between the communication interface 42 of the trust device 4 and the interface 32 of a participating device 3, may use identical or different protocols.

[0045] The enrollment device 6 comprises a processor 60, a communication interface 62 for communicating with the trust device 4 and with each participating device 3, a memory 64, a biometric sensor 66, and a document reader 68. The information provided above concerning the processor 20, the communication interface 22, the memory 24, and the biometric sensor 26 also applies to the processor 60, the communication interface 62, and the memory 64 and biometric sensor 66. The document reader 68 is typically an image sensor, such as for example a digital camera or digital camera, adapted to acquire an image of an identity document (for example a passport), and / or a contact or contactless type communication means (for example NFC, UHF, etc...) adapted to read the contents of a chip in an identity document.

[0046] In the following, we will detail an embodiment in which the enrollment device 6 is distinct from the control device 2 and the trust device 4. However, in other embodiments, the control device 2 or the trust device 4 could be used as the enrollment device 6.

[0047] Figure 2 illustrates the steps of an embodiment of a process according to the invention.

[0048] The reference biometric data may have been established in advance.

[0049] Each reference biometric data point is advantageously a data point recorded in an identity document of a reference individual. For example, the reference biometric data point may be a biometric data point obtained from a facial image appearing on an identity document (for example, a passport), or from a facial image or at least one iris of the individual recorded in a radio frequency chip contained in the identity document.

[0050] The invention uses a functional encryption for a polynomial function of degree 1 or 2 parameterized with a reference biometric data point. It is understood that there are as many different functions as there are reference biometric data points. Preferably, said polynomial function parameterized with a reference biometric data point is a "distance function with the reference biometric data point", thus allowing comparison of the function's input with the reference biometric data point, and preferably the "dot product function with the reference biometric data point", although other distance functions may be used, for example the Euclidean distance, which is of degree 2.

[0051] As will be seen, the invention is distinguished in that the input data is a vector comprising at least n proof components representing a biometric trait of a candidate individual and at least one other non-zero proof component representing a first masking element, and in that the reference biometric data comprises n reference components representing a biometric trait of a reference individual and at least one other non-zero reference component representing a second masking element, n is a natural number strictly greater than zero.

[0052] According to a first example, the input data X is a vector whose components are (xb.. .,xn,u) and the reference biometric data Y is a vector whose components are (yb.. .,yn,m) with xk the k-rank proof component representing the biometric trait of the candidate individual, u said at least one other test component representative of a first masking element (u), yk the reference component of rank k representative of the biometric trait of the reference individual, m said at least one other reference component representative of a second masking element (m).

[0053] According to a second example, the input data X is a vector whose components are (xb.. .,xn,u,v) and the reference biometric data Y is a vector whose components are (yi,.. .,yn,m,t) with xk the k-rank test component representing the biometric trait of the candidate individual, u and v two other test components representing a first masking element (u,v), yk the k-rank reference component representing the biometric trait of the reference individual, m and t two other reference components representing a second masking element (m,t).

[0054] Thus, the result s of the functional decryption of a ciphertext of the input data, also referred to hereafter as the score, is directly the result of applying to this input data the polynomial function of degree 1 or 2 parameterized with the reference biometric data, that is to say the distance of this input data with said reference biometric data in a form masked by a primary mask r. The primary mask r depends on the first masking element and the second masking element, that is to say on at least one other proof component and at least one other reference component.

[0055] The primary mask ra has the value of the result of applying to at least one other test component of the polynomial function of degree 1 or 2 parameterized with at least one other reference component.

[0056] The proposed method provides security because the distance in plain text is never calculated.

[0057] When the polynomial function of degree 1 or 2 is a dot product and when the test biometric data and the reference biometric data are defined according to the first example described above, the score has the value ,,. m , y" YY , the u tu -t- y^ primary mask r having value u 'm and the distance having value V" v .

[0058] When the polynomial function of degree 1 or 2 is a dot product and when the test biometric data and the reference biometric data are defined according to the second example described above, the score has the value u . m + v , / + . y , the primary mask r has the value 11 ' m + v ' z and the distance having the value yn YY .

[0059] It is thus sufficient to have several decryption keys (ski, sk2...), each associated with a reference biometric data, in order to be able to calculate the polynomial function for each of these reference biometric data, i.e. distances in masked form with all these reference biometric data as so many different functions.

[0060] Noting: - Dec the functional decryption function - Enc the functional encryption function - Yj, the j-eme reference biometric data, - X, a biometric proof data point, - r the primary mask, - d(A,B) the distance between A and B, - fj, the said polynomial function of degree 1 or 2 parameterized with Yj, - Why is the encryption key functional? - skj the functional decryption key for the function fj, then if the polynomial function is a distance function we have: Dec(skj, Enc(pk, X)) = fj(X) = d(X, Yj) + r.

[0061] Preferably, the decryption key(s) associated with each reference biometric data point are generated on the fly from said reference biometric data point, so that the reference biometric data point is not stored (which avoids any risk of disclosure and is also very economical in terms of storage size), and a decryption key database is created live rather than a reference biometric database. For example, each decryption key is generated directly upon reading the corresponding identity document.

[0062] In this respect, with reference to [Fig.2], the process preferably begins with an enrollment step (step E102), comprising the generation of said functional decryption key for the polynomial function of degree 1 or 2 parameterized with said reference biometric data, from a master key and said reference biometric data.

[0063] Indeed, functional encryption provides for this possibility of using a secret master key, and in the case of a polynomial function of degree 1 or 2, said functional decryption key for the polynomial function of degree 1 or 2 parameterized with said reference biometric data is typically generated by the application of said polynomial function of degree 1 or 2 between said master key and said reference biometric data, e.g. by applying a dot product between said master key and said reference biometric data if said function is the dot product.

[0064] The generation of said functional decryption key is typically implemented by the enrollment device 6 and the memory 64 of the enrollment device stores the generated decryption key. The enrollment device 6 sends the generated decryption key to the participating devices 3, typically using its communication interface 62. Each participating device 3 receives the generated decryption key from the enrollment device 6, typically using its communication interface 32, and then stores said decryption key in its memory 34.

[0065] The enrollment step (step E102) may also include the generation of said functional encryption key from the same master key, thus enabling the use of associated encryption and decryption keys. The generation of said functional encryption key is typically implemented by the control device 2.

[0066] According to another possibility, the generation of said functional encryption key is implemented by the trust device 4 and then the trust device 4 sends the generated functional encryption key to the control device 2, typically using its communication interface 42. The control device 2 receives the generated functional encryption key from the trust device 4, typically using its communication interface 22, and then stores said encryption key in its memory 24.

[0067] One can have a single master key, but alternatively one has a master key per "session", that is to say per group of identifications / authentications. For example, in the case of pre-boarding passenger screening, one has a master key per flight, which can be drawn from a list, generated randomly or using a pseudo-random algorithm, etc.

[0068] Thus, in a particularly preferred manner, the enrollment step (step E102) includes the generation of said master key, typically by the trust device 4. The master key is, for example, generated randomly or using a pseudo-random algorithm, etc. The master key is stored in memory 44.

[0069] The trust device 4 can send the master key to the enrollment device 6 and the control device 2, typically using its communication interface 42.

[0070] Also preferred, the enrollment step (step E102) includes obtaining at least one reference biometric data Yj, typically by the enrollment device 6, and, for each reference biometric data Yj obtained, generating, typically by the enrollment device 6, the functional decryption key skj for the polynomial function of degree 1 or 2 parameterized with said reference biometric data Yj, so as to constitute a functional decryption key base skj. The encryption key pk is also generated at the same time, typically by the control device 2.

[0071] The enrollment device 6 sends a generated decryption key to the devices participants 3, typically using its communication interface 62. A participant device 3 receives a decryption key from the enrollment device 6, typically using its communication interface 32, and then stores said decryption key in its memory 34.

[0072] According to another possibility, the enrollment device 6 sends the decryption key base to the participating devices 3, typically using its communication interface 62. Each participating device 3 receives the decryption key base from the enrollment device 6, typically using its communication interface 32, and then stores said decryption key base in its memory 34.

[0073] The reference biometric data Yj may not have been established in advance.

[0074] In this case, each biometric reference data can advantageously be obtained from data recorded in an identity document of a reference individual, typically reference components representative of the reference individual. For example, the reference components can be obtained from a facial image appearing on an identity document (e.g., a passport), or from a facial image or at least one iris of the individual recorded in a radio frequency chip contained in the identity document.

[0075] Obtaining at least one reference biometric data Yj can thus include, for each reference biometric data Yj, obtaining the reference components representative of the reference individual, typically by the enrollment device 6 via its document reader 68, then concatenating said reference components with the other reference components, the reference biometric data Yj being the result of this concatenation.

[0076] When reference data are established in advance, obtaining at least one biometric reference data is advantageously, for each biometric reference data, from an identity document (for example a passport), typically via the document reader 68 of the enrollment device 6.

[0077] Also preferably, the enrollment step (step E102) includes the determination by random or pseudo-random drawing of a value of at least one other component from among the at least one other reference component and the at least one other test component.

[0078] For example, u is determined by random draw and ma a predetermined value, typically 1, or vice versa.

[0079] Advantageously, when the reference biometric data are established in advance, at least one other reference component has a predetermined value.

[0080] According to another example, u and m are determined by random draw.

[0081] This determination is typically implemented by the trust device 4.

[0082] The trust device 4 can send at least one other determined reference component to the enrollment device 6, typically using its communication interface 42. The enrollment device 6 receives at least one other reference component from the trust device 4, typically using its communication interface 62.

[0083] The trust device 4 can also send at least one other proof component to the control device 2, typically using its communication interface 42. The control device 2 receives at least one other proof component from the trust device 4, typically using its communication interface 22.

[0084] When at least one other reference component and at least one other test component are determined by random or pseudo-random sampling, the security is increased.

[0085] The process itself begins with a functional encryption step (step El04) of a proof biometric data X using said functional encryption key pk.

[0086] This step is typically implemented by the control device 2.

[0087] It is important to understand that while enrollment can be implemented well before authentication / identification itself, the X proof biometric data must be obtained at the very latest a few minutes before, to ensure the "freshness" of the proof components of the proof biometric data.

[0088] Advantageously, the functional encryption step (step E104) includes obtaining said biometric proof data X from a biometric trait of the candidate individual using biometric acquisition means of system 1, typically the biometric sensor 26 of the control device 2.

[0089] As explained, the control device 2 includes a biometric sensor 26 for obtaining fresh proof components. Generally, the biometric proof data X is generated by the processor 20 from a biometric trait provided by the biometric sensor 26, typically by concatenating said proof components with other proof components, the biometric proof data X being the result of this concatenation. However, the biometric sensor 26 may include processing means for extracting the proof components and may, for example, take the form of an automatic device provided by control authorities (at the airport). Such an automatic device may, where appropriate, concatenate said proof components with other proof components.

[0090] Preferably the biometric sensor 26 is capable of detecting living beings, so as to ensure that the biometric data of test X comes from a "real" trait.

[0091] In the case where the biometric sensor 26 and the control device 2 are remote, the communication between the two can be encrypted.

[0092] The present method is distinguished, as explained in a very original way, in that the biometric proof data X is a vector comprising at least n proof components representing a biometric trait of a candidate individual, with n a natural number strictly greater than zero, and at least one other non-zero proof component representing a first masking element.

[0093] The process continues with a scoring step (step E106), during which the system 1 obtains for at least one reference biometric data Yj, a score s representing a distance between the test biometric data X and said reference biometric data Yj, by functional decryption of the encrypted test biometric data using a functional decryption key skj for a polynomial function of degree 1 or 2 parameterized with said reference biometric data Yj.

[0094] Thanks to at least one other reference component of the reference biometric data and at least one other proof component of the proof biometric data, the result of the functional encryption is not the distance of this proof biometric data X with said reference biometric data Yj in plaintext, but the distance of this proof biometric data X with said reference biometric data Yj in a form masked by the primary mask r.

[0095] The proposed method provides security because the distance in plain text is never calculated.

[0096] The process thus makes it possible to prevent an attacker from obtaining information relating to a reference biometric data without limiting the number of iterations of the process for the same reference biometric data.

[0097] The scoring step (step E106) is typically implemented by each participating device 3 of system 1. Thus, each participating device 3 obtains, for a test biometric data X and a reference biometric data Yj, the distance of this test biometric data X with this reference biometric data Yj in a form masked by the primary mask r.

[0098] According to another possibility, only one device of system 1 can implement the scoring step (step E106), typically a participating device 3.

[0099] The method may further include, for i being equal to 1 and 2, the implementation of a generation step (step E108) by a participating device 3 of index i, of a partial result o; of index i from the score s and of a demasking data k; of index i.

[0100] The participating devices 3 with respective indices 1 and 2 are distinct.

[0101] The generation step (step E108) is thus implemented by each device by- ticipant 3 of at least one pair of participating devices 3 of system 1.

[0102] The partial results of respective indices 1 and 2 allow the calculation of a control result (o) indicating whether the test biometric data corresponds or not to the reference biometric data.

[0103] The method thus allows the integration of the calculation of the distance between the test biometric data and the reference biometric data, and of the comparison of the calculated distance with a threshold, without calculation or manipulation of the distance in plain text, which further secures the method against attacks seeking to obtain information on the reference biometric data.

[0104] Furthermore, the method allows the generation of partial results without the participating devices 3 of respective indices 1 and 2 needing to exchange information between them.

[0105] If a participating device 3 of index i does not calculate the score s during the scoring step (step E106) of the process, this participating device 3 of index i can receive the score s from a device of system 1 that has calculated said score s, typically via its communication interface 32.

[0106] It should be noted that a device manipulating a score s, typically a device implementing the step of obtaining a score (step E106) and / or the step of generating a partial result (step E108), for example a device participating 3, must not have all the other proof components and all the other reference components, nor the primary mask, to ensure the security of the reference biometric data.

[0107] The unmasking data of respective indices 1 and 2 are typically generated during the enrollment step (step E102) with the primary mask r, preferably by the trust device 4, which then sends the unmasking data of index 1 to the participating device of index 1, and the unmasking data of index 2 to the participating device of index 2. For example, the primary mask r and the two unmasking data of respective indices 1 and 2 are generated by the Funshade.SetupO function described in the document "Funshade: Functional Secret Sharing for Two-Party Secure Thresholded Distance Evaluation", Alberto Ibarrondo, Hervé Chabanne and Melek Ônen, https: / / eprint.iacr.org / 2022 / 1688 / , then at least one other reference component and at least one other proof component are determined such that the primary mask ra has the value of the result of applying to at least one other proof component the polynomial function of degree 1 or 2 parameterized with at least one other reference component. Typically, when the polynomial function of degree 1 or 2 is a dot product and when the proof biometric data and the reference biometric data are defined according to the first example described above, at least one other proof component u can be determined. as follows: u = tm1, the at least one other reference component m having a predetermined value or having a value determined by random or pseudo-random draw.

[0108] It should be noted that a device manipulating a score s must not have both the unmasking data of index 1 and the unmasking data of index 2.

[0109] Preferably, the functional encryption of the biometric proof data is implemented by a device separate from the participating devices 3 of indices 1 and 2.

[0110] During the step of generating a partial result (step E108), the participating device 3 of index i calculates a partial result o; of index i from the score s and the unmasking data k; of index i: o; = FSS.eval (i,k;,s), then the participating device 3 of index i sends the partial result o; to an output device, for example the control device 2 via its communication interface 32. The function FSS.eval is typically as described in the document “Funshade: Functional Secret Sharing for Two-Party Secure Thresholded Distance Evaluation”, Alberto Ibarrondo, Hervé Chabanne and Melek Ônen, https: / / eprint.iacr.org / 2022 / 1688 / .

[0111] As previously stated, the processing consisting of the step of generating a partial result (step E108) is implemented twice: once by the participating device of index 1 and once by the participating device of index 2. Thus two partial results Oi,o2 are generated.

[0112] The process continues with a step of obtaining a control result o (step El 10), during which the system 1 calculates a control result from the partial results of respective indices 1 and 2.

[0113] The pair of partial results has the property of allowing the calculation of a control result indicating whether the test biometric data corresponds to the reference biometric data. However, it is not possible to calculate this control result based on only one of the two partial results.

[0114] The step of obtaining a control result o (step El 10) is typically implemented by an output device separate from the index devices 1 and 2, here the control device 2, which has received the pair of partial results Oi,o2 from the two participating devices 3, for example via its communication interface 22.

[0115] In one embodiment, the control result is obtained by summing the partial results, as follows: ° = °i+ °2.

[0116] Ultimately, functional encryption, followed by functional decryption, the generation of partial results, and the step of obtaining a control result allow for a comparison between a threshold and the distance between the test biometric data and the reference biometric data. The threshold is defined in the FSS.SetupO function used for generating the primary mask and the demasking data (the threshold is, in a way, encoded by this data).

[0117] In practice, the control result can be a boolean.

[0118] If the control result indicates that the biometric test data corresponds to the reference biometric data, then it is considered that the candidate individual to whom the test biometric data relates has previously been enrolled with system 1. Under these conditions, the processor 20 can command in an access control step (step El 12), an opening of the gate 28, in order to allow the candidate individual to access a secure area.

[0119] If the control result indicates that the test biometric data does not correspond to the reference biometric data, then it is considered that the individual being checked, i.e. the candidate individual, is not the reference individual to whom the reference biometric data relates.

[0120] The preceding steps (in particular the step of obtaining a score, the step of generating a partial result and the step of obtaining a control result) can be implemented U times for different decryption keys relating to different reference biometric data.

[0121] These implementations can be sequential. Alternatively, U processes can be launched in parallel, each implementing the step of obtaining a score, the step of generating a partial result and the step of obtaining a control result described above, and aggregating the final results at the access control step.

[0122] The process described above may be subject to other variations.

[0123] We can have the primary mask, at least one other proof component, at least one other reference component and the unmasking data determined once and for all, for system 1.

[0124] Alternatively, the primary mask, at least one other proof component, at least one other reference component and the unmasking data are generated, typically during the enrollment step, per "session".

[0125] According to another possibility, the primary mask, at least one other test component, and the unmasking data are single-use data for a single biometric test data item. With this possibility, the primary mask, at least one other test component, and the unmasking data can be reused multiple times, for different reference individuals, during a check of the candidate individual to whom the biometric test data relates.

[0126] Also, the primary mask, at least one other proof component and the unmasking data can be generated during the functional encryption step.

[0127] In a particularly advantageous embodiment, only a pair of participating devices 3 with respective indices 1 and 2 is used. The inventors have found that this embodiment is simple to implement, while also being reasonable It is likely fast and secure. However, it is also possible to use several pairs of participating devices implementing the process described above.

[0128] For each pair of participating devices 3, the sum of the partial results of the pair is equal to the control result. In contrast, the trusted device 4 can provide different unmasking data to different pairs of participants. The benefit of using several pairs of devices 3 is to provide redundancy to ensure that there has not been a calculation error by one pair, for example in the event of attacks.

[0129] In the process described above, each memory 34 stores a reference database which is a database of functional decryption keys.

[0130] According to one variant, no memory 34 stores the functional decryption key base, the functional decryption key base being stored in the memory 64 of the enrollment device 6. A participating device 3 which implements the scoring step (step E106) can then obtain the functional decryption key required for the execution of this step, from the enrollment device (via the communication interface 32 of said participating device and the communication interface 62 of the enrollment device) and then erase said received decryption key once the execution of the scoring step is completed.

[0131] In the foregoing, a particular application of the method was discussed, in which the result of the control determines access to a secure area. It is understood, however, that the described method can be used for other applications.

Claims

Claims

1. Method for processing biometric data, the method comprising the implementation by a system (1) of steps of: - Functional encryption (El04) of a test biometric data item (X) using a functional encryption key (pk), - For at least one reference biometric data item (Yj), obtaining a score (E106) representing a distance between the test biometric data item (X) and the reference biometric data item (Yj), by functional decryption of the encrypted test biometric data item, using a functional decryption key (skj) for a polynomial function of degree 1 or 2 parameterized with said reference biometric data item (Yj), the method being characterized in that: - The test biometric data item (X) is a vector comprising n test components (xk) representative of a biometric trait of a candidate individual, with n a natural integer strictly greater than zero, and at least one other test component (u,v) non-zero representative of a first masking element, - The reference biometric data (Yj) is another vector comprising n reference components (yk) representative of a biometric trait of a reference individual and at least one other non-zero reference component (m,t) representative of a second masking element, and - The score (s) obtained by functional decryption, represents the distance between the test biometric data (X) and the reference biometric data (Yj) in a form masked by a primary mask (r).,

2. Method according to the preceding claim in which the primary mask (r) has as its value the result of the application to the at least one other test component (u,v) of the polynomial function of degree 1 or 2 parameterized with the at least one other reference component (m,t).

3. Method according to any one of the preceding claims, further comprising determining by random or pseudo-random drawing a value of at least one other component among the at least one other reference component (m,t) and the at least one other test component (u,v).

4. Method according to any one of the preceding claims, further comprising an enrollment step (El02) comprising the generation of said functional decryption key (skj) for said polynomial function of degree 1 or 2 parameterized with said reference biometric data (Yj), from a master key and said reference biometric data (Yj).

5. Method according to the preceding claim, in which said functional decryption key (skj) for said polynomial function of degree 1 or 2 parameterized with said reference biometric data (Yj), is generated by applying said polynomial function of degree 1 or 2 between said master key and said reference biometric data.

6. Method according to any one of claims 4 and 5 in which the enrollment step (E102) comprises the generation of said functional encryption key (pk) from said master key.

7. Method according to any one of claims 4 to 6, in which the enrollment step (E102) comprises: - the generation of said master key, - the obtaining of at least one reference biometric data (Yj), and, - for each reference biometric data (Yj) obtained, the generation of the functional decryption key (skj) for said polynomial function of degree 1 or 2 parameterized with said reference biometric data (Yj), so as to constitute a functional decryption key base.

8. Method according to any one of the preceding claims, in which the functional encryption step (E104) comprises obtaining said test biometric data (X) from the biometric trait of the candidate individual using biometric acquisition means (26) of the system (1).

9. A method according to any preceding claim, wherein the polynomial function of degree 1 or 2 is a scalar product.

10. Method according to any one of the preceding claims, the method further comprising, for i being equal to 1 and 2, the implementation of the following step (E108) by a device (3) of index i of the system (1): - Generation of a partial result (o;) of index i from the score (s) and an unmasking data item (k;) of index i associated with the primary mask (r), And in which: - the devices (3) of respective indices 1 and 2 are distinct, - the partial results of respective indices 1 and 2 allow the calculation of a control result (o) indicating whether the test biometric data (X) corresponds or not to the reference biometric data (Y?-

11. Method according to the preceding claim, in which the control result (o) is equal to the sum of the partial results (ob o2) of respective indices 1 and 2.

12. Method according to any one of claims 10 or 11, in which the functional encryption of the biometric test data (X) is implemented by a device (2) of the system (1), distinct from the devices (3) of indices 1 and 2, and / or the control result (o) is calculated (step El 10) from the partial results (ob o2) of respective indices 1 and 2 by an output device (2) of the system (1), distinct from the devices (3) of indices 1 and 2.

13. A computer program comprising code instructions for executing a method according to any one of the preceding claims, when this program is executed by a system.

14. A storage medium readable by computer equipment on which a computer program comprises code instructions for executing a method according to any one of claims 1 to 14.

15. 1Z.System (1) for processing biometric data, the system being configured for implementing steps of: - Functional encryption of a biometric test data item (X) using a functional encryption key (pk), - For at least one reference biometric data item (Yj), obtaining a score (s) representing a distance between the biometric test data item (X) and the reference biometric data item (Yj), by functional decryption of the encrypted biometric test data item using a functional decryption key (skj) for a polynomial function of degree 1 or 2 parameterized with said reference biometric data item (Yj), the system being characterized in that: - The biometric test data item (X) is a vector comprising n test components (xk) representative of a biometric trait of a candidate individual, with n a natural integer strictly greater than zero, and at least one other non-zero test component (u,v) representing. indicative of a first masking element, - The reference biometric data (Yj) is another vector comprising n reference components (yk) representative of a biometric trait of a reference individual and at least one other non-zero reference component (m,t) representative of a second masking element, and - The score (s) obtained by functional decryption represents the distance between the test biometric data (X) and the reference biometric data (Yj) in a form masked by a primary mask (r).

Citation Information

Patent Citations

  • METHOD AND SYSTEM FOR PROCESSING BIOMETRIC DATA

    FR3123130A1

  • Verification of biometric templates for privacy preserving authentication

    US20220131698A1