GB2544087A

GB2544087A - Automatic teller machine with biometric authentication

Info

Publication number: GB2544087A
Application number: GB1519590.2A
Authority: GB
Inventors: Condon Adrian
Original assignee: B-Secur Ltd
Current assignee: B-Secur Ltd
Priority date: 2015-11-05
Filing date: 2015-11-05
Publication date: 2017-05-10
Also published as: GB201519590D0

Abstract

An automatic teller machine (ATM) system, 200, with biometric authentication comprising, a bio-potential waveform sensor, 204, coupled to a biometric authentication unit, 210. The biometric authentication unit is adapted to compare a bio-potential waveform of a user with a pre-recorded reference bio-potential waveform associated with said user. The bio-potential waveform may be an electrocardiogram (ECG, EKG) pulse. The reference waveform may be stored on a user card, 209. The bio-potential waveform may be on a contactless card or a smart phone. The waveform sensors may be integrated into a key/button of the ATM. There may be two or more sensors and at least one on each side of the keypad, 201. The user may be informed to place a finger/digit onto its own sensor, 213. The sensors may be integrated into the display/screen, 207, of the ATM. A second inventive concept is provided for modifying/retrofitting an existing Automated Teller Machine with the bio-potential waveform sensors.

Description

Automatic Teller Machine with Biometric Authentication Field of the Invention

The invention relates to the integration of biometric authentication in automatic teller machines. Background of the Invention

Biometric measurements can include, but are not limited to, methods such as facial recognition, iris recognition, vein patterns, DNA sequencing and fingerprint scanning, as well identification through personal characteristics such as vocal properties, gait properties, handwriting features and many more. These biometric measurements have found use in many applications, including security access, user identification and criminological identification. Problems have however been encountered with biometric authentication systems, such as inaccuracy, ease of forgery and poor discrimination between individuals which has led to only a limited use of these technologies in the marketplace.

User authentication on an automatic teller machine (ATM) is typically carried out through inputting a 4 digit personal identification number (PIN); this however is not a completely secure means by which to authenticate an ATM user; instances of fraud through techniques such as card skimming or card cloning are widespread, and the PIN of a user can easily be acquired. A need exists for a more secure method of user authentication at an ATM, and could be addressed through a suitably secure biometric authentication system. An ATM system with biometric authentication, using bio-impedance measurement of a user's hand, has previously been suggested in W02000/016245.

Bio-potential waveforms such as those from electrocardiograms (ECG) are generated by the electrical activity of an individual's heart and vary from person to person based upon variations in heart morphology, orientation, ion paths, blockages etc. Bio-potential waveforms have been identified as offering the ability for biometric personal identification. An algorithm for human recognition from ECG signals has previously been disclosed in "ECG Based Recognition Using Second Order Statistics", F. Agrafioti and D. Hatzinakos, Communication Networks and Services Research Conference, pp. 82-87, 2008.

Summary of the Invention

The inventors have devised novel and inventive automatic teller machine systems with biometric enrolment and authentication and methods for biometric enrolment and authentication of automatic teller machine transactions. A broad description will be given of specific aspects of the invention. Preferred features of the specific aspects are set out in the dependent claims.

This invention uses a bio-potential waveform authentication system for application in ATMs.

In one aspect, an ATM is fitted with bio-potential waveform sensor electrodes, which are integrated into one or more keys of the ATM. These keys are used to measure an ECG waveform of the user. The user has a conventional chip and pin or smart card upon which a reference ECG waveform for the user is stored during an enrolment process. A biometric unit or module is included in the ATM which is used for authentication of the user by comparing the measured ECG waveform and a reference ECG waveform of the user. The bio-potential waveform preferably comprises at least one electrocardiogram (ECG) pulse.

In another aspect, the bio-potential waveform sensor electrodes are located on a mobile phone or dedicated module (such as a smart card or key fob) which communicates wirelessly with the ATM. Preferably, in this case, a comparison between the actual ECG signal from the user and the prestored ECG signal is performed in the smart phone or other module and this permits authentication between the smart phone (or other module) and the ATM (or a server connected to the ATM) thereby uniquely identifying the user to the ATM for dispensing cash or for another transaction. A method for the operation of the ATM with biometric authentication is detailed, whereby a user places one or more digits on one or more sensors of the ATM, an ECG waveform of the user is recorded, this is compared to a reference ECG waveform of the user stored upon a chip and pin or smart card he or she has inserted into the ATM. A biometric unit compares the measured and reference ECG waveforms and either approves or denies the user access to continue their ATM transaction. A method for retrofitting an existing ATM is detailed. This method involves the removal of one, two or more standard keys from a keypad of an ATM and the replacement with one or more keys in which ECG bio-potential waveform sensor electrodes are integrated. Further to this, a biometric pre-processing circuit is preferably also installed in the existing ATM. Biometric authentication can be performed by an existing ATM processor or by an additional processor provided for that purpose or by a smart card.

This invention solves existing problems in ATM security. In the standard ATM operating system a user inserts his or her card and enters a PIN corresponding to the card; this PIN can easily be acquired used by another person who may have stolen or cloned the user's card. Other biometric authentication systems can also be breached, for example in a fingerprint scanning biometric system the victim could have his or her finger or thumbprint copied, or even have their finger or thumb removed with the offender then using this to gain access to the victim's bank account. An ECG waveform overcomes these issues, as intrinsically an ECG waveform is internal to the user and cannot be stolen from the user and therefore only the specified user can gain access to his or her bank account.

The invention also represented an extremely convenient means for performing ATM transactions, as it permits secure identification of a user without any card or other hardware.

Brief Description of the Drawings

Fig. 1 shows a plan view of the keypad, in which sensors for biometric authentication are integrated.

Fig. 2 shows a conceptual view of an ATM system according to a preferred embodiment of the present invention.

Fig. 3 shows a block diagram of a biometric authentication process according to an aspect of the present invention.

Fig. 4 shows a plan view of the keypad, in which sensors for biometric authentication are integrated, according to an alternative embodiment of the invention.

Fig. 5 shows a networked system according to aspects of the present invention.

Detailed Description of the Invention

Figure 1 shows a representation of an ATM keypad (101) comprised of sixteen keys separated into a numerical keypad (102) comprised of twelve keys and a secondary keypad (103) of four keys used for operational processes such as confirming or clearing an entry. As will be apparent to one of skill in the art, the present invention can be applied to other arrangements of keypad.

The twelve keys of the numerical keypad (102) are numbered from 0 to 9 and can be used, for example, for the user to enter a personal identification number (PIN) or specify the amount of cash he or she wishes to withdraw. In the embodiment displayed in Figure 1 the two keys of the numerical keypad, positioned to the left and right of the "0" key have electrodes integrated (104) upon which the user can place two digits (105) (fingers or thumbs) for biometric authentication.

In Figure 2 an ATM (200) is shown, where a display unit (205) and keypad (201) are coupled to a CPU (211) , along with a card slot (208) into which can be inserted a card (209) associated with a user. Such a user card can be a credit card, a debit card, a cash card, or a chip-and-pin card. In some aspects, the user card is a smart card, which includes an integrated circuit and may include contactless functionality. Figure 2 further shows a biometric unit (210), and a cash dispensing means (212) . The ATM may include a magnetic strip reader, chip reader or other means to read information, such as user identification information or a user's reference waveform, from the card. Of course, when the user card is a smart card with contactless functionality, there is no requirement for a card slot, but instead, the user card may be held against a designated area of the ATM and the contactless functionality utilised to read information from the card.

In the arrangement shown in Figure 2, the display unit (205) comprises a display screen (207) with four display keys (206) located to the left and right of the display screen (207). The keypad (201) contains two banks of keys; a numerical keypad (202) and a secondary keypad (203). The keys immediately to the left and right of the "0" key have bio-potential waveform sensors installed (204) which are connected to the biometric unit (210). In other arrangements, the display unit can comprise any other number of keys, or the display unit may have no keys.

In a preferred embodiment, the bio-potential waveform sensor keys (204) are designed to be integrated into the ATM keypad (201) as, for example, replacements for at least two of the other keys of the keypad (201). A bio-potential waveform sensor key (204) is a key, such as an ATM key, or other sensor pad area, that has incorporated therein a bio-potential waveform sensor (or biopotential waveform electrode). In some aspects, the bio-potential waveform sensor keys (204) can be integrated into existing keys such that the keys still maintain their original functionality but also act as sensors (electrodes) for biometric authentication. In some aspects, the two or more biopotential waveform sensors (204) can be located on the ATM but not on the keypad. For example, the sensors (204) may be located either side of the ATM keypad (101). In other aspects, the sensors (204) are located on the display unit (205), for example replacing two or more of the display keys (206). In some aspects, two or more bio-potential waveform sensors may be integrated into a single bio-potential waveform sensor key.

The bio-potential waveform sensor keys (204) may be fitted as standard in new ATM machines during the manufacturing process, or said keys (204) may be retrofitted to existing ATM machines by substitution of the original keys, or original keys may be modified to include bio-potential waveform sensors (electrodes). The sensors (electrodes) in the bio-potential sensor keys (204) are preferably made from a conductive material, preferably gold plate, but other materials such as doped semiconductors, ceramics, silver or copper may be used, as well as any other conducting material.

Preferably the material of the electrode has an IACS rating of at least 70 (meaning at least 70% the conductivity of copper) but, in air, is more resistant to surface oxidation than copper (e.g. has a lower rate of surface oxidation).

In some aspects the, for example gold plate, may cover the entire surface of the key. In other aspects the conductive material may partially cover the surface of the keys in a patterned formation. Patterning of the electrode on the key (e.g. in tracks) could allow for less conducting material to be required and thus act as a cost saving and anti-theft measure if the conducting material is a precious metal.

The bio-potential waveform sensor electrodes (204) are connected to a biometric unit (210). This may be a mere signal pre-processing module (e.g. amplification and filtering) or may include enrolment and/or authentication functionality. An existing ATM may be modified, or retrofitted, to include a biometric unit (210). The biometric unit can be fitted at the same time as the bio-potential waveform sensor electrode keys (204) or before or after fitting the bio-potential waveform sensor electrode keys (204). In other aspects, a biometric unit (210) may be fitted during the manufacturing of a new ATM. In further aspects the biometric enrolment and/or authentication processing may be carried out by an existing processor in an ATM, such as the central processing unit of the ATM.

The biometric unit (210) and/or the CPU (211) preferably perform heartbeat detection, e.g. QRS or PQRST detection, for example using a Pan-Tompkins algorithm.

Figure 3 shows a block diagram detailing an operation process of the ATM with biometric authentication according to aspects of the invention, on the assumption that a user is pre-enrolled for the biometric authentication process. Enrolment is described below..

The user approaches the ATM and (optionally) presents his or her user card wirelessly or inserted into the card slot (301). A reference bio-potential waveform of the user is stored on the user card. (An entirely cardless process is described below.) The user is prompted to initiate the biometric authentication process (302) by placing two fingers (or other digits, such as thumbs) onto the keys in which electrodes are integrated (303). Figure 2 shows an example in which a user prompt is displayed (213).

Preferably the user will place a digit, where a digit is defined as one finger or thumb, from each hand onto the electrodes, as is illustrated in Figure 1, as this will provide a bio-potential ECG waveform with a strong signal by measuring the potential difference across the heart.

Fleartbeat detection is carried out and, if no heartbeat is detected (e.g/ within a timeout period), the CPU (211) causes a prompt to be displayed on the display (207) to prompt the user to try again, e.g. to wipe his or her fingers or to try different fingers. When at least one heartbeat has been detected, the process proceeds to step (304).

Multiple waveforms can be obtained from the user during a single biometric authentication process. Obtaining multiple waveforms increases accuracy and reduces the risk of a spurious read, which itself could lead to a false negative or false positive when authenticating the user.

The bio-potential waveform is monitored (304) and the authentication process compares the recorded bio-potential waveform with the reference bio-potential waveform of the user stored on the user card (305).

The authentication process preferably comprises (a) windowing, where a pre-processed ECG signal is subjected to segmentation into non-overlapping windows, (b) normalized autocorrelation computation for every window and (c) dimensionality reduction (e.g. by Discrete Cosine Transform or Linear Discriminant Analysis). Details of these steps are given in the above paper by Agrafioti. An authentication decision is taken based on comparison of the ECG signal and a pre-stored ECG signal.

If the biometric authentication process approves, the user they may then access the account associated with the inserted card and continue their ATM transaction (306).

Preferably the keys into which the sensors (electrodes) are integrated (204) are on the left and right sides of the keypad (201) to encourage the user to place a digit (finger or thumb) from each hand onto each sensor in order to provide the strongest possible ECG signal. The arrangement of the keys into which the electrodes are integrated is however not restricted to this. Any two keys of the numerical keypad (202), the secondary keypad (203) or the display keypad (206) may have the sensors (electrodes) integrated with no restriction as to which side of the keypad (201) or display (205) they are located. In some aspects, more than two keys of the numerical keypad (202), the secondary keypad (203) or the display keypad (206) may have the sensors (electrodes) integrated. Alternatively, a single key can be used, with two electrodes on the key, as it remains possible to measure a satisfactory biopotential waveform of an ECG nature from just one finger.

The embodiment of the invention illustrated in Figure 1 shows the keys to left and right of the "0" key being used as the keys into which the electrodes are integrated (104). In this embodiment these keys are chosen as these keys are present on many ATM systems but do not have a numerical input value assigned to them, they are therefore well suited to having the electrodes integrated without causing confusion to the user. This arrangement is however only preferable and does not limit the system to only using these two keys; any two keys on any of the numerical keypad (102, 202), secondary keypad (103, 203) or the display keypad (206) may have the electrodes integrated.

As noted above, an existing ATM can be retro-fitted (modified or adapted) with the appropriate means to perform biometric authentication. For example bio-potential waveform sensors (204) can be integrated into the keys on the ATM. This is advantageous compared to building new sensors into an ATM, as it allows for the retrofitting of existing ATMs to include the bio-potential waveform sensors, thereby reducing the need to replace and dispose of existing ATMs. By simply removing at least two standard buttons from an existing ATM and installing at least two bio-potential waveform sensor keys the ATM can be upgraded for biometric authentication using ECG waveforms. At a minimum, these electrodes can be connected to spare inputs (e.g. spare analog-to-digital inputs) on an existing ATM CPU.

When retrofitting the bio-potential waveform sensor keys (204), a biometric unit (210) may also be installed to prepare the signal from the electrodes for on-site biometric authentication at the ATM. This biometric unit may be coupled to the CPU of the ATM for operational purposes and the CPU may perform authentication by suitable software update. This retrofitting process offers the advantage of being a faster and lower cost method of upgrading an ATM compared to removing the entire machine and replacing it with one containing biometric authentication means. In some aspects, the existing CPU of the ATM may be completely replaced with an upgraded CPU that performs the function of the existing CPU, the biometric pre-processing means and the authentication.

Figure 4 shows an alternative embodiment of the invention. An ATM keypad is shown (401), consisting of a numerical keypad (402) and a secondary keypad (403). In this embodiment of the invention, the biometric authentication process is carried out by placing two fingers, or a finger and thumb, of the same hand (405) on the two bio-potential waveform sensor electrode keys (404). In Figure 4 the sensors (electrodes) are integrated in the leftmost key of the bottom row of the keypad and the "0" key. The biometric authentication process is then carried out as previously detailed. This approach may not provide such a strong bio-potential waveform signal as the approach in which one digit from each hand is used, however it offers the ability for a disabled user, such as a user with only one hand, to operate the ATM.

In some aspects, as shown in Figure 5, part of the biometric processing is carried out at a server of the user's bank (503). For example, the user may not need to use a card at all, instead the user may begin the authentication process at the ATM (501) by simply applying two digits to the electrode sensors built into the ATM keys, their ECG waveform would then be recorded and transmitted through a network (502) such as the Internet to a server, such as bank server (503), at which is stored a reference ECG waveform of the user. The measured ECG waveform may then be compared to the reference ECG waveform to identify the user. Once the user is identified, a permission signal is sent from the bank server (503), through the network (502) and back to the ATM (501) where the user is granted permission to continue his or her transaction. If the user is not identified, a denial signal can be sent to the ATM, which can then inform the user that the request has been denied.

The advantage of this embodiment is that if, for example, the user's card has been lost or stolen the user would still be able to access his or her account and withdraw emergency cash.

In some aspects, the user may identify themselves to the ATM (501), by presenting or inserting a user card or by inputting a personal identification number, for example. The ATM (501) can then send a request for a reference waveform to the server (503), via the network (502), along with identification information. The server (503) can then extract the reference waveform from a database, based on the identification information. The server (503) then transmits the reference waveform to the ATM (501), which compares the reference waveform with a waveform obtained from the user via the bio-potential waveform sensors at the ATM (501) for verification of the user.

In some aspects of the authentication process it may be that the user inserts his or her user card to the ATM (501), he or she is then prompted to have his or her ECG waveform recorded, and then both the recorded waveform and the reference ECG waveform from the user card are transferred securely via a network (502) to the server (503). At this server (503) the biometric authentication process is executed and if approved the bank server sends a permission signal back through the network to the ATM allowing the user to continue their transaction. This embodiment provides the advantage that less retrofitting to the ATM is required, as a biometric authentication processor does not need to be installed.

In some aspects the user is not require to use a user card at all. For example the user may utilise an app (e.g. iBeacon™ - https://developer.apple.com/ibeacon/) or near field communication (NFC) on a smart phone that has its own biometric potential waveform electrodes. In this case, the reference ECG waveform may be stored at the server (503) and comparison may be made at the server. Alternatively, the reference ECG waveform is stored on the smart phone. In this case, the comparison can optionally be at the server or on the smart phone. For example, the ATM can present a challenge to the smart phone that calls for a response, but the response can only be released to the ATM if there is a satisfactory biometric potential waveform authentication.

Alternatively, authentication is undertaken at the ATM using the CPU (211) by comparison of the presented biometric potential waveform against the biometric potential waveform stored on the phone and sent to the ATM.

An additional feature of this aspect is that the ATM can send a message to the phone to indicate it is nearby and available for this service.

In some aspects the biometric authentication process is an added feature to the existing approach of using a user card at an ATM, e.g. if biometric authentication fails, the user can instead use a standard PIN authentication. Alternatgively, the two can be used together for greater security.

The user can enrol for biometric authentication at the ATM.

The user identifies himself or herself to the ATM by some means acceptable to the server. This may be by presentation of a smart card, using chip-and-pin authentication, or by secret passcode or challenge-and-response etc. The user requests enrolment (e.g. by choosing an option in a menu) and is prompted to present his or her digit(s) to the appropriate sensor(s). The ATM measures the biopotential waveform and conveys this (e.g. in encrypted form) to the server 503. Thereafter, the server uniquely associates that biopotential waveform with that user. Thereafter, the user can present himself/herself to another ATM and perform the authentication process described above.

Alternatively, on enrolment, the user presents a card (e.g. a smart card) and the reference biopotential waveform of the user is stored on the user card.

By integrating the sensors into the existing buttons of an ATM the user may enter a personal identification number, or may use the ECG process if he/she is already enrolled or wishes to enrol.

Many other variants and embodiments will be apparent to the skilled reader, all of which are intended to fall within the scope of the invention whether or not covered by the claims as filed. Protection is sought for any and all novel subject matter and combinations thereof disclosed herein.

Claims

1. An automatic teller machine (ATM) system with biometric authentication comprising; at least one bio-potential waveform sensor coupled to a biometric authentication unit, the or each bio-potential waveform sensor being integrated into an ATM, wherein the biometric authentication unit is adapted to compare a bio-potential electrocardiogram (ECG) waveform of a user with a pre-recorded reference bio-potential waveform associated with said user.

2. The system of claim 1 wherein said ATM comprises a card slot, adapted to receive a user card having stored thereon the reference bio-potential waveform, and means to read the reference bio-potential waveform from said card.

3. The system claim 1 wherein said ATM comprises a wireless transceiver, adapted to communicate with a contactless card, smart phone or other wireless device to receive a biopotential waveform therefrom.

4. The system of any preceding claim where one bio-potential waveform sensors is integrated in a key on the left side of the ATM keypad and another of the bio-potential waveform sensors is integrated in a key on the right side of the ATM keypad.

5. The system of any preceding claim where the bio-potential waveform sensors are positioned on the lower leftmost and rightmost keys of the ATM keypad.

6. The system of any preceding claim wherein the ATM comprises a display unit.

7. The system of claim 6 wherein the display unit is adapted to prompt the user to place an indicated digit or indicated digits on corresponding bio-potential waveform sensors on the ATM.

8. The system of claim 6 wherein the display unit is adapted to prompt the user to place at least two digits from either the same or both of the user's hands on at least two biopotential waveform sensors on the ATM.

9. The system of any of claim 6 to 8, wherein the bio-potential waveform sensors are integrated into a display unit of the ATM.

10. The system of any preceding claim where the bio-potential waveform sensors are gold plated.

11. The system of any preceding claim where the bio-potential waveform sensors are coupled to a biometric unit.

12. The system of any preceding claim wherein the biometric unit is coupled to a central processing unit of the ATM.

13. The system of claim 1 to 2 or 4 to 12 wherein a server stores the reference bio-potential waveform that is accessed at the ATM through a network.

14. A method for the biometric authentication of an automatic teller machine (ATM) transaction comprising; obtaining, via at least two bio-potential waveform sensors, a bio-potential waveform from of the user; comparing the obtained bio-potential waveform to a reference bio-potential waveform of the user; authenticating the user if the obtained bio-potential waveform is in agreement with the reference bio-potential waveform; and permitting an ATM transaction based on the authentication.

15. The method of claim 14 where the bio-potential waveform comprises at least one ECG pulse.

16. The method of claim 14 or 14 further comprising reading the reference bio-potential waveform from a user card associated with the user in advance of comparing the obtained bio-potential waveform to a reference bio-potential waveform.

17. The method of any of claims 14 to 16 wherein obtaining a bio-potential waveform of the user comprises the user placing at least two digits and at least one digit from each of the user's two hands on at least two bio-potential waveform sensors on the ATM.

18. The method of any of claims 14 to 16 wherein obtaining a bio-potential waveform of the user comprises the user placing at least two digits from either the same or both of the user's hands on at least two bio-potential waveform sensors on the ATM.

19. The method of any of claims 14 to 18 further comprising transmitting the obtained biopotential waveform to a server via a network in advance of comparing the obtained biopotential waveform with the reference bio-potential waveform.

20. The method of any of claims 14 to 18 further comprising transmitting identification information of the user to a server via a network and receiving a reference bio-potential waveform in advance of comparing the obtained bio-potential waveform with the reference bio-potential waveform.

21. The method of any of claims 14 to 20 wherein the at least two bio-potential waveform sensors used for bio-potential waveform signal detection are integrated into at least two of the ATM keys.

22. The method of any of claims 14 to 21 further comprising prompting a user to place two or more digits on the at least two bio-potential waveform sensors.

23. A method for modifying an ATM for biometric authentication comprising; installing one, two or more ATM keys with bio-potential waveform sensors; and coupling the bio-potential waveform sensors to the central processing unit of the ATM directly, or indirectly through additional pre-processing circuitry.

24. The method of claim 23 where two or more existing keys from an ATM are removed and replaced with keys comprising conducting sensors for bio-potential waveform signal detection.

25. The method of claim 23 where the ATM keypad is replaced with a modified keypad comprising keys with integrated conducting sensors for bio-potential waveform signal detection.

26. A method for biometric enrolment at an automatic teller machine (ATM) comprising; uniquely identifying a user; prompting the user to present one or more digits to sensors on the machine; obtaining a bio-potential waveform from the user; storing a representation of the bio-potential waveform; and uniquely associating the bio-potential waveform with the identified user.