Patent Application
20130198851
This invention relates to online transactions, and more particularly, to ways to help secure sensitive data during online transactions.
Online transactions such as purchase transactions often require that entities such as merchants and payment card processors exchange sensitive information. For example, in connection with a typical purchase by a customer, a merchant often obtains the primary account number (PAN) corresponding to the payment card account of a customer (e.g., the customer's credit card number). The merchant then provides the PAN to a payment card processor (payment processor) as part of an authorization request. The payment processor then uses a tokenization server to obtain a corresponding token from a central database that is provided to the merchant if the purchase is authorized.
Later, when settling the purchase transaction, the merchant may submit the token and the settlement amount to the payment processor. The payment processor may recover the PAN of the customer from the token using the central database.
In this type of transaction, regional tokenization servers typically process purchase transactions with merchants in a given geographical region. Regional tokenization servers typically request the token corresponding to the provided PAN from a common central database. If no token corresponding token exists in the common central database at the time of the request, a new token is generated and an entry is added to the common central database.
Purchase transactions originating with merchants around the world are therefore dependent on global communications systems to provide access to the common database. This can be problematic if any portion of a global communications system becomes temporarily unavailable or inoperable.
In some situations, regional tokenization servers can generate partially tokenized sensitive data at a tokenization server based on one or more previously provided token look-up tables without relying on constant connection to a global communications network. However, these systems typically tokenize only a portion of the sensitive data due to data storage and processing considerations.
It would therefore be desirable to be able to provide improved ways in which to handle sensitive data such as tokens in connection with online transactions.
Online transactions may involve exchanges of sensitive information between entities and organizations. It may be desirable for an entity such as a merchant or a government agency to protect the sensitive information from exposure (e.g., by maintaining access to the sensitive information without storing the sensitive information).
An entity may protect sensitive information such as a sensitive number without storing the sensitive information by storing a token that corresponds to the sensitive information. The token corresponding to the sensitive information may be acquired by sending a token request that includes the sensitive information from the entity to a token generating organization.
A token generating organization may include geographically diverse tokenization equipment. Blocks of tokenization equipment may be located in multiple cities or other geographic locations. Each city (or other geographic location) may be associated with separate tokenization equipment.
A token may be generated by a tokenization server on one of multiple tokenization systems associated with tokenization equipment in a particular geographic location. Each tokenization system may be implemented on a physically distinct hardware platform. Each hardware platform may include physically separate computing equipment such as a physically a separate power supply, one or more processors, storage and/or other computing equipment.
Each tokenization system may include a tokenization server and an associated database. A tokenization server may generate the token corresponding to the sensitive information using a fractional token table stored on the associated local database. Each local database at each tokenization system associated with the token generating organization may be provided with the fractional token table.
The fractional token table may be pre-populated with partial tokens and corresponding partial sensitive numbers. The tokenization table may be used to map portions of the sensitive information such as portions of the sensitive number to partial (fractional) tokens.
A tokenization server may generate a full token corresponding to sensitive information such as a full sensitive number based on the fractional token table. For example, the tokenization server may perform a predetermined number of rounds of a Feistel network to generate a full token corresponding to a full sensitive number using partial tokens and a partial sensitive numbers stored in the fractional token table. A fractional token table may, for example, be a half-token table (i.e., a table having half-tokens and uniquely corresponding half-sensitive-numbers).
Once the token has been generated by the tokenization server, the token may be provided to the token requesting entity (token requestor). The token may be temporarily stored at the token requestor. The sensitive information may later be recovered by sending a request from the token requestor to the token generating organization for the sensitive information. The request may include the token and additional information such as a monetary settlement amount associated with a purchase transaction.
A tokenization server at one of the tokenization systems associated with the token generating organization may recover the full sensitive number from the full token using the fractional token table in the local database at that tokenization system. For example, the tokenization server may perform the same predetermined number of rounds of the Feistel network in reverse to recover the full sensitive number from the full token using the partial tokens and the partial sensitive numbers stored in the fractional token table.
Sensitive information may include a sensitive number such as a social security number, driver license number, primary account number associated with a credit card, or other sensitive information. Token requesting entities involved in protecting sensitive data for online transactions such as purchase transactions may be merchants (e.g., hotels, online merchants, grocery stores, etc.), universities, government agencies or other entities that use or process sensitive information. Token generating organizations may include banks, credit card companies, government agencies, corporations, payment card processors or other organizations having tokenization systems provided with fractional token tables.
Further features of the invention, its nature and various advantages will be more apparent from the accompanying drawings and the following detailed description of the preferred embodiments.
A token generating organization may generate tokens corresponding to sensitive information in order to protect the sensitive information for online transactions.
Entities that process online transactions may obtain sensitive information associated with an online transaction. It may be desirable for an entity to retain access to the sensitive information for a period of time without risking exposure of the sensitive information.
Entities may obtain sensitive information that includes a sensitive number (SN) from a person, a customer, an organization, etc. A sensitive number may be a social security number (SSN), a credit card number (CCN), a license number, or other private number associated with a person or organization. Entities may desire to maintain access to, for example, a primary account number (PAN) associated with a payment card without risking exposure of the PAN by storing the PAN.
Entities may, for example, be merchants that authorize and settle payment card purchase transactions by communicating electronically with organizations such as payment card processors. A conventional system 10 in which payment card transactions are often authorized and settled is shown in
Clients 12 commonly submit an authorization request including a customer's credit card number (CCN) to a tokenization server 18 over network 16 to determine whether the customer is authorized to make a purchase. In response to receiving an authorization request, a tokenization server 18 typically returns a token (TOKEN) over network 16 to the client that corresponds uniquely to the customer's PAN.
Tokenization servers 18 are commonly connected over a portion of communications network 16 (i.e., portion 16D of network 16) to a common central database 20. Common database 20 maintains a list 22 of primary account numbers (PANs) and corresponding tokens. Each entry in common database 20 includes a PAN (i.e., CCN1, CCN2, etc.) and a unique token for that PAN (i.e., TOKEN1, TOKEN2, etc.). When presented with a PAN, tokenization server 18 consults database 20 to determine whether an entry for that PAN already exists. If an entry is present, the tokenization server may retrieve the token associated with the PAN. If no entry is currently present for the PAN, tokenization server 18 generates a token for the PAN and stores the generated token in database 20 for future use.
After obtaining a token for the PAN from central database 20, tokenization server 18 returns the token to merchant 12 for use in settling the purchase transaction.
At a later time, the client settles the purchase transaction by submitting a settlement request to tokenization server 18. The settlement request includes the token and the final purchase transaction amount (i.e., a monetary value). In processing the settlement request, tokenization server 18 retrieves the PAN associated with the transaction based on the token from database 20 over network 16D. After the settlement request has been processed the client typically is informed of successful completion of the purchase transaction over network 16.
Conventional systems such as system 10 of
It may be desirable to distribute list 22 to regional databases for regional access that is not dependent on communications network 16. However, regional databases having local copies of list 22 may be problematic. For example, if multiple databases are maintained, a credit card number may become associated with more than one token. In another example, a single token may become associated with more than one credit card number resulting in confusion over who is responsible for payment related to a transaction.
These problems may be avoided by pre-populating a list on each and every regional database that contains a unique token for each and every possible credit card number. However, because sensitive numbers such as credit card numbers typically contain as many as, for example, 9, 12, or 16 digits, the required size of a list of all possible numbers is impractically large for storage and searching even for modern computing systems.
It may therefore be desirable to provide multiple distributed tokenization systems with a pre-populated fractional token table containing partial tokens and partial sensitive numbers along with equipment configured to generate a unique full token corresponding to any full sensitive number using the pre-populated fractional token table.
An illustrative token generating organization 30 having distributed tokenization systems at which tokens may be generated using pre-populated fractional token tables and provided to token requestors is shown in
Token requesting entities such as token requestor 32 may be brick-and-mortar merchants (e.g., hotels, stores, etc.), may be online merchants, may be universities, government agencies, or other entities. Token generating organization 30 may be associated with one or more payment card companies, banks, credit card companies, government agencies, hotels, etc. For example, token generating organization 30 may be affiliated with a bank or a credit card company or may be a third party service that services payment requests for multiple credit card companies (as examples). As an example, token requestors 32 may be properties of a hotel chain that while token generating organization 30 may be a token generating organization owned and operation by the hotel chain.
Communications network 35 may include wired and wireless networks, local area networks, wide area networks such as the Internet, or other suitable communications paths. As shown in
Tokenization equipment 37 may include geographically diverse equipment. For example, token generating organization 30 may have offices in multiple cities or other geographic locations. Each city (or other geographic location) may be associated with separate tokenization equipment. For example, equipment 1 may be located in a first city and equipment 2 may be located in a second city. Hardware redundancy may also be provided at each of these locations. For example, equipment 1 may have a first hardware platform 34 and a separate redundant hardware platform 34. The first and second hardware platforms (e.g., first and second tokenization systems implemented on first and second separate computers respectively) may be co-located. For example, the first and second computers may be located in a common building in the first city. Geographic redundancy may be provided using one or more hardware platforms at one or more additional geographic locations (see, e.g., tokenization equipment 2, etc.).
With this type of arrangement, there are multiple computers (hardware platforms) each of which forms a respective substantially similar redundant tokenization system for token generating organization 30. During operation, token requestors 32 may submit token requests to organization 30 over a communications network. Load balancing equipment such as distributor 35D may route incoming requests among systems 34 to ensure that no single one of systems 34 becomes overburdened with requests.
Token requests may include sensitive information such as a sensitive number (SN). Token requests including sensitive numbers may be sent from each token requestor 32 and may be routed by distributor 35D to an appropriate tokenization system 34 (e.g., in tokenization equipment 1 or tokenization equipment 2 or a particular system 34 at either of these locations) based on current loading at each tokenization system 34, based on geographical proximity of token requestor 32 to particular tokenization equipment 37, based on which of two or more redundant co-located systems 34 is in operation at a particular geographic location and has available capacity, based on the availability of portions of network 35 between token requestor 32 and a particular tokenization system 34 or based on other considerations.
For example, a token request from token requestor 32 may be routed to tokenization equipment 1. During routing of the token request to tokenization equipment 1, distributor 35D may receive information that tokenization equipment 1 is unavailable due to network disruptions in a portion of network 35 between distributor 35D and tokenization equipment 1. In response to receiving information that tokenization equipment 1 is unavailable, distributor 35D may route the token request to tokenization equipment 2. If desired, tokenization equipment such as tokenization equipment 1 may handle token requests from token requestors 32 in a given associated geographic region (e.g., a continent, a country, multiple countries, a state, multiple states, a city, multiple cities, parts of a city, or other geographic region). However, this is merely illustrative. Each tokenization system 34 may be configured to handle token requests from token requestors 32 in any geographic region. Other types of load partitioning may be used, if desired (e.g., to route requests based on request type, time of day, or other routing criteria).
Token generating organization 30 may include any number (e.g., one, two, three, more than three, four, more than four, etc.) of tokenization systems 34. As shown in
Each hardware platform may be implemented on physically distinct computing equipment including physically separate power supplies, processors, storage, etc. (i.e., separate computers or separate racks of computing equipment, etc.). Each hardware platform 34 may use its computing equipment to execute code associated with encryption and decryption processes, database management processes, authentication processes, settlement processes, token generation processes, sensitive information recovery processes, fraud detection processes, etc.
The use of multiple physically distinct hardware platforms (tokenization systems) at each geographic location such as hardware platforms A and B in equipment 1 may help ensure that organization 30 has the ability to generate tokens at a particular location (e.g., in the event that one hardware platform loses power or otherwise becomes inoperable). The use of hardware platforms (tokenization systems) at geographically diverse locations helps guard against regional service disruptions.
Each hardware platform 34 may include a tokenization server 36 and an associated database 38. Each database 38 may be provided with a common fractional tokenization table such as fractional token table 40. Each tokenization server 36 may be configured to generate a token corresponding to sensitive information in a token request by accessing fraction token table 40 on the associated database 38.
In the example of
In this example, hardware platform A of tokenization equipment 1 includes tokenization server T1-A coupled to associated database D1-A, hardware platform B of tokenization equipment 1 includes tokenization server T1-B coupled to associated database D1-B, hardware platform A of tokenization equipment 2 includes tokenization server T2-A coupled to associated database D2-A, and hardware platform B of tokenization equipment 2 includes tokenization server T2-B coupled to associated database D2-B. However, this is merely illustrative. Each block of tokenization equipment 37 may include any number of hardware platforms and each token generating organization 30 may include any number of regional blocks of tokenization equipment 37.
Fractional token table 40 of database D1-A of tokenization equipment 1 may be a common tokenization table that is identical to fractional token table 40 of databases D1-B, D2-A, D2-B, etc.
Any tokenization server 36 of a token generating organization 30 may be configured to generate a common full token that corresponds uniquely to any full sensitive number received in a token request using common fractional token table 40. For example, in response a token request containing a particular sensitive number, tokenization servers T1-A and T2-B may generate a same token corresponding to that particular sensitive number. In this way, a single sensitive number such as a credit card number may correspond uniquely to one and only one token and a single token may correspond uniquely to one and only one sensitive number, regardless of which tokenization server generates the token.
Tokenization table 40 may be pre-populated with partial sensitive numbers such as partial social security numbers, partial license numbers, partial credit card numbers, etc. In one embodiment which is sometimes described herein as an example, sensitive information in a token request may include a primary account number (PAN) associated with a payment card. In this example, tokenization table 40 may include partial primary account numbers (PANs) (e.g., half-PANs, quarter-PANs, eighth-PANs or other fractional PANs) and uniquely corresponding partial tokens (e.g., half-tokens, quarter-tokens, eighth-tokens, or other respective fractional tokens).
A token requestor 32 may desire to obtain a token corresponding to sensitive information obtained at the token requestor. For example, when a customer desires to make a purchase at a merchant, the merchant obtains a PAN associated with a payment card account of the customer (e.g., the customer's credit card number). As an example, a magnetic card reader or other computing equipment at the merchant may be used to read the PAN from the customer's card.
A token requestor such as token requestor 32 may submit a token request that includes the sensitive information to token generating organization 30 over communications network 35. The token request may be routed by distributor 35D to appropriate tokenization equipment 37.
In response to receiving a token request, the tokenization equipment 37 may generate and return a token that corresponds uniquely to the sensitive information in the token request to token requestor 32.
For example, the merchant may submit an authorization request to a token generating organization such as a payment processor to determine whether the customer is authorized to make a purchase. The authorization request generally includes the PAN and a requested authorization amount. The authorization amount may be somewhat larger than the actual expected purchase price. The payment card processor may determine whether the customer is authorized to make a purchase of the requested amount. If authorized, the payment processor may generate and return a token to the merchant that corresponds uniquely to the customer's PAN.
Each entry in fractional token table 40 may include a partial sensitive number and a unique partial token for that partial sensitive number. When presented with a sensitive number such as a social security number, a PAN, etc. associated with sensitive information in a token request, tokenization server 36 may generate a full token corresponding to the full sensitive number using the fractional token table 40 on the local database 38.
A sensitive number associated with sensitive information in a token request may, for example, contain 16 digits. Fractional token tables 40 stored on databases 38 of systems 34 may, for example, be half-token tables that contain a list of each and every possible 8 digit number (e.g., 0000 0000 through 9999 9999). A half-token table that includes a list of all possible 8 digit numbers may include a corresponding list of 8 digit tokens, each uniquely corresponding to one of the 8 digit numbers in the half-token table.
Tokenization servers 36 may be configured to use hardware platform computing equipment to generate full tokens corresponding uniquely to full sensitive numbers by running a predetermined number of rounds of a tokenization cycle. Each tokenization cycle may include tokenizing portions of a sensitive number (e.g., a PAN, a SSN, etc.) and modifying the tokenized portions of sensitive number. In the example of a half-token table, tokenization servers 36 may be configured to tokenize a first half of a sensitive number and to generate a full token corresponding to the full sensitive number in part by modifying the tokenized first half of the sensitive number using a non-tokenized second half of the sensitive number. The step of modifying the tokenized first half of the sensitive number may be implemented as part of a round of a Feistel network run on the tokenization server to generate the full token using the fractional token table.
After generating a full token for the sensitive number in the token request with tokenization server 36, tokenization system 34 may return the full token to token requestor 32.
At a later time, token requestor 32 may want to recover the sensitive number (e.g., to recover a PAN in order to settle the purchase transaction). The token requestor may submit a sensitive-information-recovery (SIR) request to token generating organization 30. The SIR request may include the full token and, if desired, additional information (e.g., a monetary value such as the final purchase transaction amount). In the example of a purchase transaction involving a token, a final purchase transaction amount may be equal to a previously authorized amount or may be different from a previously authorized amount. As an example, if an authorization was requested for $5000, the final purchase transaction amount may be for $1034.
A sensitive-information-recovery request may be routed by distributor 35D to any of tokenization equipment 37 of token generating organization 30. One of tokenization servers 36 at the tokenization equipment 37 that receives the settlement request may recover the full sensitive number (e.g., a full PAN, a full SSN, etc.) from the full token using the fractional token table 40 (e.g., a half-token table) stored on the corresponding local database 38. Tokenization server 36 may recover the full sensitive number from the full token using the fractional token table by running the tokenization cycle in reverse for the same predetermined number of rounds that were used to generate the full token.
In the example of
Before providing fractional token table 40 to databases 38, fractional token table 40 may be populated with partial tokens such as half-tokens 42 generated using a random number generator (e.g., a pseudo-random number generator, a truly random number generator, a hardware-generated random number generator, etc.). If desired, partial tokens such as half-tokens 42 may be shuffled with respect to half-SNs 44 (e.g., using a Knuth shuffle or other shuffling process).
In a typical token request, a full token corresponding to a full SN such as a full PAN may be desired. Full tokens may be generated by tokenization servers such as tokenization servers 36 of
A diagram showing an example of a suitable method for tokenizing partial sensitive numbers such as half-SNs and modifying and combining tokenized partial sensitive numbers and un-tokenized partial sensitive numbers to generate full tokens is shown in
A tokenization cycle such as tokenization cycle 50 may be run on a chosen one of tokenization servers 36 for a predetermined number of rounds. Tokenization cycle 50 may include running a full sensitive number such as full SN 52 through a Feistel network for the predetermined number of rounds by first separating full SN 52 into a first half such as left half L and a second half such as right half R. This is merely illustrative. If desired, any two portions of full SN 52 may be used in tokenization cycle 50.
In the example of
The resulting left half-token L′ from the reversible addition at junction 56 (i.e., the addition of left half L and the tokenized right half RT of full SN 52) may then become left half-token L′ of full token 54. The result L′ may also be tokenized using half-token table 40 (e.g., L′ may be treated as a half-SN and mapped to a uniquely corresponding half-token) to generate a tokenized half-token LT'. The tokenized result LT′ may then be added to right half R of full SN 52 as indicated by junction 58 to produce a right half-token R′=R+LT'. As described above in connection with addition 56, tokenized half-token LT′ may be added to right half R at junction 58 using a reversible addition process (e.g., an addition process such as digit-wise addition, arithmetic addition with truncation, an XOR process, or other addition process that, when run in reverse by subtracting the tokenized half-token LT′ from right half-token R′ recovers right half R of full SN 52).
Tokenization cycle 50 may be repeated for a predetermined number of rounds (e.g., one round, two rounds, three rounds, four rounds, more than four rounds, 10 rounds, more than ten rounds, 100 rounds, more than 100 rounds, etc.) by running the Feistel network of
Full token 54 may be formed by suitably combining the left half-token L′ and right half-token R′. In the example of
A full sensitive number such as full SN 52 may be recovered from full token 54 by running the Feistel network of tokenization cycle 50 in reverse (for the same number of predetermined rounds that were used in generating full token 54), as shown in
For example, a merchant may submit a SIR request such as a settlement request to a token generating organization such as a payment processor that includes the full token (and, if desired) the final monetary purchase transaction amount. Tokenization servers 36 at tokenization equipment 37 may be configured to recover full SN 52 from full token 54 using a fractional token table such as half-token table (
The result R of the subtraction at junction 60 (i.e., R=R′−LT′) may then become the recovered right half R of full SN 52. The result R may also be tokenized using half-token table 40 to generate tokenized right half RT. Tokenized right half RT may then be subtracted from left half-token L′ of full token 54 as indicated by junction 62 to recover left half L of full SN 52. As described above in connection with subtraction 60, tokenized right half RT may be subtracted from left half-token L′ at junction 62 using a subtraction process that reverses the addition performed at junction 56 (e.g., a subtraction process in which the result of the subtraction at junction 60 is left half L of full SN 52).
In configurations in which tokenization cycle 50 was run for a predetermined number M rounds to generate full token 54, full SN 52 may be recovered from full token 54 by running tokenization cycle 50 in reverse (as described in connection with
As described above in connection with
As shown in
The addition of each digit of numbers 70 and 72 results in a number (as indicated by array of numbers 74) that may be equal to, larger or smaller than 9. Numbers smaller than or equal to nine may form a digit of result 76. The first digit of numbers larger than nine may be dropped so that only the trailing digit is used to form a digit of result 76. The digit-wise addition process illustrated by
If desired, a reversible addition process at junctions 56 and 58 may include a length-preserving arithmetic addition as shown in
In the example of
If desired, a reversible addition process at junctions 56 and 58 may include an exclusive OR (XOR) operation as shown in
Binary result 90B of the XOR operation may then be converted into a decimal representation to generate result 90. The XOR addition process illustrated by
Illustrative steps involved protecting sensitive data in online transactions by generating tokens for token requestors using tokenization servers implemented on tokenization equipment of a token generating organization are shown in
At step 100, a fractional token table such as half-token table 40 of
At step 102, a token request that includes a full SN may be received by tokenization equipment of token generating organization 30 over a communications network such as communications network 35 from a token requestor. A token request may, for example be included in an authorization request associated with a purchase transaction at a merchant.
At step 104, a full token that uniquely corresponds to the full SN in the token request may be generated using a tokenization server such as a selected one of tokenization servers 36 of tokenization equipment 37 that received the token request. The full token may be generated from the full SN by performing a predetermined number M rounds of a tokenization cycle. Each round of the tokenization cycle may include a round of a Feistel network that uses a fractional token table stored on the local database of the tokenization system. A full token may be generated by combining (e.g., appending) two or more fractional tokens (e.g., half-tokens). A fractional token may be generated in part by tokenizing a portion of the SN and modifying the tokenized portion of the SN with at least one reversible addition.
At step 106, the full token may be provided to the token requestor from the tokenization equipment.
The steps shown in
Illustrative steps involved in recovering a sensitive number corresponding to a token received by tokenization equipment in a sensitive-information-recovery request from a token requestor are shown in
At step 110, a sensitive-information-recovery request that includes the full token that was provided by a tokenization generating organization to the token requestor may be received from the token requestor at any of tokenization equipment 37. The SIR request may be received at the same tokenization system that generated the token or may be received at a different tokenization system.
At step 112, the full SN may be recovered from the full token by performing the predetermined number M rounds of the tokenization cycle in reverse. Each round of a tokenization cycle performed in reverse may include a round of a Feistel network run in reverse that uses the fractional token table stored on the local database of the regional payment processor. The full SN may be recovered by combining (e.g., appending) two or more fractional SNs (e.g., half-SNs). A fractional SN may be recovered in part by tokenizing a portion of the full token and modifying the tokenized portion of the full token with at least one subtraction operation that reverses a reversible addition operation performed in generating the full token from the full SN.
Some SIR requests such as SIR requests involving social security numbers or other identifying numbers may require that the sensitive information (e.g., the sensitive number) be returned to the token requestor. Other SIR requests such as settlement requests associated with purchase transactions involving payment cards may be processed (e.g., settled) at the tokenization equipment without returning the sensitive information to the token requestor.
If the SIR request does not require the sensitive information (e.g., the sensitive number) to be returned to the token requestor, a tokenization system may proceed to step 114. At step 114, the tokenization system computing equipment may be used to finalize the transaction using the full SN recovered from the full token at the tokenization system.
If the SIR request requires the sensitive information (e.g., the sensitive number) to be returned to the token requestor, a tokenization system may proceed to step 116. At step 116, the tokenization system computing equipment may be used to provide the full SN recovered from the full token at the tokenization system to the token requestor.
The foregoing is merely illustrative of the principles of this invention and various modifications can be made by those skilled in the art without departing from the scope and spirit of the invention.
This application claims the benefit of provisional patent application No. 61/592,536, filed Jan. 30, 2012, which is hereby incorporated by reference herein in its entirety.
| Number | Date | Country | |
|---|---|---|---|
| 61592536 | Jan 2012 | US |
