WO2007134417A1

WO2007134417A1 - Method for identifying a typing motor pattern

Info

Publication number: WO2007134417A1
Application number: PCT/BR2007/000124
Authority: WO
Inventors: André Luiz Tadeu Alves SILVA
Original assignee: Idéias Seguras Tecnologia E Consultoria Em Softwares Ltda
Priority date: 2006-05-24
Filing date: 2007-05-22
Publication date: 2007-11-29

Abstract

The method comprises the detection of parameters of time intervals when an individual is typing a reference set of characters (CR) on a keyboard. The typing of the reference set of characters (CR) allows the calculation of collection averages (MC) of the parameters of each typing of the reference set, of a final average (MF) of the useful collection averages MC and of an item average (MI), by dividing the final average (MF) by each collection average (MC), the highest and lowest values of the item averages (MI) being used as reference limits for the patterns obtained in subsequent typings. The values of the collection average MC, obtained in each typing operation subsequent to an initial number 'n' to detect the motor pattern are incorporated to the calculation of a new final average (MF) and of new item averages (MI), adjusting the pattern to the characteristics developed by the individual.

Description

METHOD FOR IDENTIFYING A TYPING MOTOR PATTERN Field of the Invention

The present invention refers to a method for identifying an individual using a keyboard, such as a keyboard of a computer, a telephone, a touch-screen, a mouse or any other terminal device, to be used in different applications such as, for example, to electronically access an institution or a system of restricted access, by typing a code or password on said keyboard. Prior Art

There are well known different prior art methods for providing identification of a user of a keyboard system, in which the user is identified by his typing characteristics when inputting a certain code, password or message in the keyboard. Some of these known methods are described in documents US 4,621,334, US 5,557,686, US 6,151,593 and US 2004/0059950 Al.

However, these known methods propose the formation, in different manners presenting higher or lower reliability, of a motor pattern of typing a code, password or phrase in a keyboard, by an individual to be identified. Said individual typing pattern is recorded in a database and associated with that individual, to be used as a "reference signature" in the validation of subsequent operations, such as those in which the individual electronically accesses the institution or system of restricted access.

The methods described in the first three documents cited above include the elaboration of a typing motor pattern by analyzing the typing characteristics of a user to be identified when typing a determined set of characters

(generally letters and/or numbers) . The typing motor pattern can be identified since the beginning, from a certain number of samples, including all the data collected in each sample or only those considered appropriate. These methods present the inconvenience of considering the thus obtained typing motor pattern as being the pattern of that individual, independently of the physical abilities presented by the individual in different occasions, mainly as a function of the increase, with time, of its abilities to type the characters that define the code, password or reference phrase for the measurement of the typing times characteristic of the individual. There is not foreseen any automatic update of the typing motor pattern of the individual with time. This inconvenience does not guarantee the reliability of the detected pattern, permitting the occurrence of frequent errors in the identification of the individual who requires access to the institution or system. Document US 2004/0059950 describes a method for determining a reliable pattern, including a phase in which the individual learns how to type a reference set of characters and a phase in which the individual has his pattern determined with the use of only some collected samples, considered representative of a pattern in relation to the typing of individual keys. The pattern is determined by the individual typing motor patterns selected and the user of the institution or system is compelled to remain within that pattern associated to him. Despite the higher level of reliability of the typing pattern associated with each user, this previous method does not provide the automatic update of the pattern during the time in which the individual remains as a user of the institution or system to be accessed. Disclosure of the Invention As a function of the facts mentioned above, it is an object of the present invention to provide a method for identifying a typing motor pattern, i.e., a "digital handwriting", which can be easily implemented in different equipment already being used and which can be aggregated in any programming language, allowing the progressive and automatic identification and update of the typing motor pattern, without interference of the user and while the latter is effecting the usual personal identification operations, such as those to electronically access the institution or system of restricted access . It is a further object of the present invention to provide a method as mentioned above, which allows using any set of characters that can be typed as a reference element to form the typing motor pattern, without requiring prior knowledge or motor ability from the user in relation to said reference set of characters.

The method proposed by the invention concerns the identification of a motor pattern of an individual upon typing a predetermined reference set of characters on a keyboard which, as already mentioned, may be defined as a keyboard of a computer, a telephone, a touch-screen, a mouse, or any other terminal device.

According to the invention, the method comprises the steps of : a) requiring the individual to type said reference characters on the keyboard; b) detecting a parameter defined by at least one time interval of pressing and release of the keyboard upon typing the reference characters; c) calculating a collection average for the detected parameter and storing it in a database; d) repeating the steps "a" , "b" and "c" for a certain "n" number of times that is predetermined as a function of the security level to be obtained; e) calculating a final average of the collection averages of "n" typing operations of the reference characters, utilizing only the collection averages whose values are situated within a maximum acceptable deviation in relation to said final average; f) calculating an item average for each one of the ^ΛΛn" operations of typing the set of characters, by dividing the final average by each one of the respective collection averages; g) recording, in a database: as "upper limit", the highest value and, as "lower limit" , the lowest value of the item averages of the "n" operations of typing the set of characters; h) repeating the steps "a" , "b" and "c" in a subsequent operation, of number n+1, of typing the reference characters by an individual ; i) calculating the item average for said operation of number "n+1", dividing the final average, already stored in the database, by the new collection average ; j ) verifying whether the new item average of the operation of number "n+1" is contained between the "upper and lower limits" of the item averages recorded in the database, to validate the identification of the typing motor pattern of said operation of number "n+1" ; k) after validating the new item average, recalculating the upper and lower limits of the "n" item averages, discarding the first item average and considering the one relative to the operation of number "n+1" ; 1) recording, in the database: as upper limit, the highest value and, as lower limit, the lowest value of the item averages of the last "n" operations of typing the set of characters; and m) repeating the steps from "h" to "1" for each new operation of typing the set of characters by an individual on said keyboard.

As already mentioned, the present method is particularly adequate to provide a security identification of an individual, for example, a user of an institution, such as a financial institution, or a user of a system whose access, by electronic means and through a keyboard, is restricted, requiring validation of identification upon an access operation.

With the methodology defined above, the invention permits, in a relatively fast and simple manner, the formation of a typing motor pattern of an individual in relation to a predetermined set of characters, which can be defined by an alphanumeric password or by a phrase of easy memorization.

The method for identifying a "digital handwriting" object of the present invention further presents, as a positive aspect, the possibility of being integrated to any system, independently of the language in which it has been described, since it is an algorithm that can be written in the form of a function or an object, not being obligatory the use of a specific software in an institution or system, which restricted access is validated through the identification of the typing motor pattern of a user when he wants to access said protected institution or system. The function or object related to the utilized algorithm can be activated in parallel to the running of the system at any time, having its running in background, not interfering in the usual and routine activities of an institution or system at the exact moment of verifying the identity of a user or collecting data to define the typing motor pattern. Brief Description of the Drawings

The present invention will be described below, with reference to the enclosed drawings, given by way of example of an embodiment of the invention and in which: Figure 1 represents a diagram showing how the time intervals are obtained: between the pressing and the release of the keyboard; between the release of the keyboard, after typing each character, and the pressing of the keyboard for typing the subsequent character; and between the pressing of the keyboard for typing the first character and the release of the keyboard after typing the last reference character;

Figure 2 represents a flowchart of the method steps during a procedure of forming the typing motor pattern of an individual; and Figure 3 represents a flowchart illustrating the method steps during a procedure of validating an individual by identifying his typing motor pattern in the predetermined reference set of characters. Detailed Description of the Invention

As previously mentioned, the present method is particularly used for validating the identification of users of protected institutions and systems, which electronic access is only permitted upon the correct identification of the individual requesting said access. Nevertheless, it should be understood that the present method may be utilized in different applications which require the personal identification of an individual through his motor pattern of typing a set of characters. As mentioned above, the present method utilizes the identification of a motor pattern of typing on a keyboard, which can be a physical keyboard, such as the keyboard of a computer, a telephone or any other terminal device, or even a virtual keyboard that is operated, for example, through a mouse or a touch-screen. The present method utilizes a preliminary technique for elaborating a personal typing motor pattern, automatically updating said pattern at each operation of typing the reference set of characters .

In order to effect the identification of the typing motor pattern of an individual, there is preferably predetermined a reference set of characters CR defined by a number preferably greater than three and which can consist of different types of characters of the type usually found in physical or virtual keyboards, these reference characters being generally defined by alphanumeric characters. In the case of an application for identifying the typing motor pattern to allow the electronic access to the institutions or systems of restricted access, the reference set of characters can consist of, for example, the numeric, alphabetic or alphanumeric password that is supplied to the user by the institution or system to be accessed. It should be understood that these reference characters can present the most different configurations, as long as they permit the identification of the elements to be pressed on a physical or virtual keyboard, so that the individual may execute the typing of the characters that constitute the reference set, whose typing motor pattern must be identified by the present method. Figure 1 illustrates the use of a reference set of characters CR formed by four characters.

For carrying out the identification method, as illustrated in figures 1 and 2, the individual is initially required to type, on the keyboard, said reference characters CR previously indicated to him and which generally consist of reference characters unique to each individual . Subsequently, the method includes a step of detecting a parameter defined by at least one time interval To, Ti, Tt of pressing and release of said keyboard, while typing said reference characters CR.

In a way of carrying out the invention, the parameter detected upon the pressing and release of the keyboard is defined by the time intervals To between the pressing and the release of the keyboard while typing each reference character CR.

In another way of carrying out the invention, said parameter to be detected in the pressing and the release of the keyboard is defined by the time intervals Ti between the release of the keyboard, after typing each reference character CR, and the pressing of the keyboard for typing each respective subsequent reference character CR. Yet another way of carrying out the invention allows the parameter, which will be detected in the pressing and in the release of the keyboard, to be defined by the total time interval Tt elapsed between the pressing of the keyboard, for typing the first reference character CR, and the release of the keyboard after typing the last reference character CR. The three ways of determining said parameter to be detected in the pressing and in the release of the keyboard can be utilized separately or in conjunction, the utilization depending on the level of sophistication and security to be imparted to the method as a function of the target public.

Although the third way suggested above for defining the parameter to be detected in the pressing and in the release of the keyboard is used without the other two ways, it should be understood that it is preferably applied always jointly with the first or the second way of determining the parameter, in order to increase the verifying elements of the identification method. The procedures mentioned above for defining the reference set of characters CR and determining the parameters related to the pressing and the release of the keyboard are represented in the flowchart of figure 2 through the first two blocks. Thus, phase 10 of forming the reference set of characters is followed by a phase in which the individual is instructed, in a instruction phase 20, to sequentially type, on an adequate keyboard, as defined above, the reference set of characters CR. While typing the reference characters, there is detected said parameter representative of the motor characteristics of the individual upon the pressing and the release of the keyboard while typing each one of the reference characters CR.

The identification method further includes a verifying phase 30, in which it is verified whether characters typed by the individual correspond to the predetermined characters of the reference set of characters CR. In case of discrepancy, the individual is instructed to re-type the reference characters CR.

It should be understood that the present method is carried out in conjunction with processing devices, not illustrated, associated with devices, such as a keyboard, a display, etc., to execute the present identification method, and which may be also associated with an institution or system of restricted access . Accordingly, the parameters related to the motor characteristics of typing the reference set of characters CR are duly stored to be subsequently processed in a calculation phase or step 40 (see figure 2) , to determine a collection average MC for the parameter detected in each operation of typing the reference characters CR and in which each collection average MC is stored in a database (not illustrated) . The calculation of the collection average MC, for each reference set of characters CR typed by the individual, is obtained by dividing the already obtained time interval parameter by the number of its ocurrences. Thus, considering time intervals To or Ti related to the times of pressing and release of the keyboard, in the typing of the different reference characters CR, the collection average MC for typing each reference set of characters CR will be the result of the sum of the different parameters represented by said time intervals, divided by the number of time intervals To or Ti utilized. MC = (To₁ + To₂ + To₃ + ...To_z)/z MC = (Ti₁ + Ti₂ + Ti₃ + ... Ti₂.) /z MC = Tt

In the formulas above, "z" represents the number of parameters detected in each typing of the set of characters. The number "z" is equal to the quantity "q" of characters of the reference set of characters, when the parameters collected are defined by the time intervals To between the pressing and the release of the keyboard upon typing each reference character CR. It should be understood that "z" is equal to "q-1" when the parameters collected are defined by the time intervals Ti between the release of the keyboard, after typing each reference character CR, and the pressing of the keyboard for typing each subsequent respective reference character CR.

After calculating the collection average MC relative to the typing of a first reference set of characters CR, the method comprises a phase of obtaining pattern 50, in which occurs the repetition of the steps of typing, on the keyboard, said reference characters RC, for determining the parameters related to the motor characteristics of said typing and further determining a new collection average MC, for each one of these operations that are subsequent to the first operation of typing the reference set of characters CR. These operations can be carried out automatically, for example, when the individual accesses an institution or system of restricted access, without the need of requiring him to execute any additional procedure other than the simple typing of the reference set of characters CR. While said initial operations are carried out, as represented by block 50 in figure 2, the number of collection averages MC will progressively increase until reaching a certain number "n" of collections, each presenting a collection average MC calculated as mentioned above. The number "n" of initial samples used for determining the typing motor pattern of an individual can vary depending on the level of reliability to be imparted by the method and required by the application to which it is directed. One may imagine, for example, a number "n" equal to 25, or even greater, as being sufficient to determine a typing motor pattern or a preliminary "digital handwriting" for an individual, as commented ahead.

In the example of the pattern forming phase illustrated in figure 2, while the collection averages MC for typing each reference set of characters are being determined from two collection averages MC, the method can include a verifying step or phase 60, according to which it is verified whether the collection averages being obtained are situated within a maximum acceptable deviation in relation to a determined final average MF, to be described ahead. Accordingly, and as illustrated in figure 2, while the subsequent collection averages MC are being determined with the successive typings of the reference set of characters CR by the individual, the method includes the periodic and progressive calculation of a final average MF of the collection averages MC already determined as a function of the already completed typings, allowing those collection averages MC, whose values are situated out of a maximum acceptable deviation in relation to said final average MF, to be discarded. In this case, the number ^λN' of collections or collection averages MC, to be utilized for determining said final average MF, must be maintained, requiring the individual to effect new typings of the reference set of characters CR, to calculate the final average MF, until the desired "n" collection averages MC, situated within a predetermined maximum deviation in relation to the final average MF, are useful to calculate said final average MF:

MF = (MCi + MC₂ + MC₃ + MC_n) /n The procedure mentioned above, in which "n" can be any integer from 2 onwards, allows the typing pattern, which will be identified, not to be verified, taking into account anomalous operational conditions, as in the cases in which, for example, the individual is drunk, irritated, sleepy or in an unusual condition. Momentary deviations during the typing operation are also ignored, which deviations can be caused by, for example, a telephone call, sneeze, distraction, etc.

Thus, the final average MF of the collection averages MC can be formed, during the process of typing the reference set of characters CR, for example, already upon determining the second collection average MC, each collection average MC being subsequently determined and incorporated in the calculation of the final average MF to be finally utilized for determining the reference pattern of the individual.

In a way of carrying out the invention, the maximum acceptable deviation for the collection averages MC is obtained by comparing the reference between the highest and lowest values of the collection averages MC with the final average MF and discarding the collection averages MC which present a difference greater than the final average MF value obtained.

After determining the useful "n" collection averages MC, the method includes a calculation step or phase 70, in which the final average MF is calculated from the useful collection averages MC in "n" operations of typing the reference characters CR.

After calculating the final average MF for the "n" collection averages MC considered useful and representative of the current pattern of the individual, the method provides a step of forming limits 80, in which an item average MI is calculated for each of the useful "n" operations of typing the reference set of characters CR. The calculation of the item averages MI is made by dividing the final average MF by each of the collection averages MC used in the formation of the final average MF, i.e., MI= MF/MC, recording in a database, as represented by the block 90, the highest value and the lowest value of the item averages MI as elements that define an upper limit and a lower limit, respectively, for said item averages MI , whereby the phase for the preliminary determination of the typing motor pattern of an individual is completed.

From the determination of the motor pattern of typing the reference set of characters CR, subsequent operations of validating the identification of an individual can be made, by verifying the timing characteristics of typing the reference set of characters CR.

For validating the identification of an individual, the user is requested to execute a new operation of a number n+1 of typing the reference set of characters CR, as illustrated in figure 3, in block 20a, in order to obtain the parameter of time interval To or Ti or Tt and calculate a respective collection average MC for said typing operation n+1 of the reference set of characters CR, in the calculation phase represented by the block 40a.

The calculation phase 70a is then executed, in which the final average MF is calculated, which corresponds to the collection average MC of the operation n+1, when said collection average MC is useful, i. e., presents an acceptable deviation in relation to the final average MF already stored in the database, during the phase of defining the typing motor pattern.

After calculating the final average MF for the typing operation of number n+1, the phase of calculating the item average MI 80a for said typing operation is initiated, by dividing the available final average MF by the new collection average MC and, then, the validation operation in the phase 100 (see figure 3) is completed, verifying whether the item average MI obtained from the typing operation "n+1" is contained within the upper and lower limits stored in the database. In case the new item average MI is out of the upper and lower limits, the validation is rejected, and a new typing is requested, returning to the phase of reference characters CR.

On the other hand, when the new item average MI is within the upper and lower limits, the latter are recalculated, by inserting the value of the new collection average MC, relative to the typing operation n+1, in the calculation of a new final average MF, discarding the first collection average MC and the corresponding first collection item average MI, recording the new limits in the database, as indicated in block 110 and validating the identification of the individual, as indicated in block 120.

As it can be noted, the present method permits the typing motor pattern of each individual to be automatically and easily obtained upon completion of an adequate, variable, but predetermined number of operations of typing a predetermined reference set of characters, which operations can follow a frequency compatible with the usual and periodic procedures required for the periodic identification of the user in different applications. Besides the aspect cited above, the present method allows the detected pattern to be automatically adjusted at each subsequent operation of typing the set of characters by an individual, since at each new operation, the parameters obtained in said operation are automatically incorporated in the calculation of the reference indices to be utilized in the identification of the current pattern of each individual .

As it can be noted, the present method can utilize different comparative mathematical procedures, giving to the institutions, which are potential users of the method, exclusivity in the methodology of calculating the reference values, which are indicative of the motor pattern of each individual upon typing the reference set of characters CR related to said individual . This characteristic of the proposed solution makes irrelevant, for the security of the system which is utilizing the method, the leak of information concerning the institution and the reference set of characters. The present method is based on overlapping linear averages, and it can be configured and calibrated according to the requirements of the institution.

The calculation of the item average MI for each operation of typing the reference set of characters CR is made according to the following function, in which the time interval parameter is defined by the time interval To between the pressing and the release of the keyboard, upon typing each reference character RC.

MT =^■ MF

MC,

_{MF ,}(MC₁₊MC₂₊...MCj n (MC₁ +MC₂ +..MC_n)

MI = n

MC_x (MC)I →n

Ml^' =^■ rø

MC.

where :

X = number of a given collection or operation of typing the reference set of characters; qCR = quantity of reference characters in the reference set of characters and which, in the example given, corresponds to the number "z" of collected parameters; n = number of operations of typing the reference set of characters;

RC = reference characters; MC = collection average

MI = item average

To = time interval between the pressing and the release of the keyboard, upon typing each reference character.

By sampling the function X, the highest and the lowest interval of the sample are defined, utilizing the concept of function limits.

For the limit, we have:

Lim f (X) = L

X A Where A is a real number different from f (x) and greater than 0 (zero). That is, A f (x) / a > 0.

Thus, the surjective function f (x) will be defined according to two limits given by the equation above, having its inversion carried out to determine the decreasing, minor limit.

There are simpler manners for defining limits between not defined intervals. The matrices, arrays, are one of the methods utilized. The mathematical example proposed herein is only for didactic and explanatory purposes to facilitate the understanding of the algorithm as a whole. After defining the limits, it is possible to make an interval range to determine whether determined collection or sample is indeed reliable and pertains to the authorized user, the same user who has registered his collection data.

The intervals can be calibrated, utilizing only the higher and lower limits, henceforth respectively denominated A and B.

1 type: A is less than X which is less than B.

2 type: A is less than or equal to X which is less than or equal to B .

3 _>rd type: A is less than or equal to X is less than B.

4^th type: A is less than X which is less than or equal to

B.

While the present invention has been described using only some exemplary embodiments, it should be understood that alterations in the means utilized can be made, without departing from the inventive concept of the method for identifying a typing motor pattern, i.e., a "digital handwriting" , as defined in the claims that accompany the present specification.

Claims

1. Method for identifying a typing motor pattern of a predetermined reference set of characters, by an individual on a keyboard, characterized in that it comprises the steps of: a) requiring the individual to type said reference characters (CR) on the keyboard; b) detecting a parameter defined by at least one time interval of pressing and release of the keyboard upon typing the reference characters (CR) ; c) determining a collection average (MC) for the time interval parameter detected upon the typing of the reference characters (CR) and storing it in a database; d) repeating steps "a" , "b" and "c" for a certain ^ΛΛn" number of times that is predetermined as a function of the security level to be obtained; e) calculating a final average (MF) of the collection averages (MC) of "n" typing operations of the reference characters (CR) , utilizing only the collection averages (MC) whose values are situated within a maximum acceptable deviation in relation to said final average (MF) ; f) calculating an item average (MI) for each one of the "n" operations of typing the set of characters, by dividing the final average (MF) by each one of the respective collection averages (MC) ; g) recording, in a database: as upper limit, the highest value and, as "lower limit" , the lowest value of the item averages (MI) of the "n" operations of typing the set of characters; h) repeating steps "a" , "b" and "c" in a subsequent operation, of number n+1, of typing the reference characters (CR) by an individual; i) calculating the item average (MI) for said operation of number "n+1", by dividing the final average (MF), already stored in the database, by the new collection average (MC) ; j) verifying whether the new item average (MI) of the operation of number "n+1" is contained between the upper and lower limits of the item averages (MI) recorded in the database, to validate the identification of the typing motor pattern of said operation of number "n+1" ; k) after validating the new item average (MI) , recalculating the upper and lower limits of the "n" item averages (MI) , discarding the first item average (MI) and considering the one relative to the operation of number "n+1";

1) recording, in the database: as upper limit, the highest value and, as lower limit, the lowest value of the item averages (MI) of the last "n" operations of typing the set of characters; and m) repeating the steps from "h" to "1" for each new operation of typing the set of characters by an individual on said keyboard.

2. Method, as set forth in claim 1, characterized in that the parameter detected in the pressing and release of the keyboard is defined by the time intervals (To) between the pressing and the release of the keyboard upon typing each reference character.

3. Method, as set forth in claim 1, characterized in that the parameter detected in the pressing and release of the keyboard is defined by the time intervals (Ti) between the release of the keyboard, after typing each reference character, and the pressing of the keyboard for typing each subsequent respective character.

4. Method, as set forth in claim 1, characterized in that the parameter detected in the pressing and release of the keyboard is defined by the total time interval (Tt) elapsed between the pressing of the keyboard, for typing the first reference character, and the release of the keyboard after typing the last reference character.

5. Method, as set forth in claim 1, characterized in that the maximum acceptable deviation for the collection averages (MC) is defined in relation to the final average (MF) of the collection averages (MC) already determined during the typing of the reference set of characters (CR) .

6. Method, as set forth in claim 5, characterized in that the final average (MF) of the collection averages (MC) is formed during the typing of the reference set of characters, from the determination of the second collection average (MC) , each subsequently determined collection average (MC) being incorporated in the calculation of the final average (MF) .

7. Method, as set forth in claim 5, characterized in that the maximum acceptable deviation for the collection averages (MC) is obtained by comparing the difference between the highest and lowest values of the collection averages (MC) with the final average (MF) and discarding the collection averages (MC) which present a difference greater than the value of the final average (MF) .

8. Method, as set forth in claim 1, characterized in that it includes, in step "g" , the additional step of calculating the difference between the highest and the lowest value of the item averages (MI) , so as to make it available as an additional parameter for identifying a typing motor pattern.

9. Method, as set forth in claim 1, characterized in that the reference set of characters comprises characters usually contained in a keyboard of physical or virtual characteristics .

10. Method, as set forth in claim 9, characterized in that the reference characters comprise alphabetic and /or alphanumeric characters.

11. Method, as set forth in claim 1, characterized in that the reference set of characters is defined by a particular code or password of an individual, to be typed on a keyboard, in order to validate the electronic access to an institution or system of restricted access.