Student Name: Kathlee een n Silveira ra Student ID: 420004 0440 - - PowerPoint PPT Presentation

Student Name: Kathlee een n Silveira ra Student ID: 420004 0440 Supervisor Name: Mehmet t Orgun un

• 1. Introduction
• 2. Functional Design
• 3. Technical Design
• 4. Evaluation
• 5. Conclusion

2

• What is Biometrics?
• What is Keystroke Dynamics?
• Aims, Outcomes and Significance
• Methodology
• Past Work

3

4

 Bio (life) and metric

(to measure)

• Biometric is a method

to measure certain human characteristics

 unique to a person,

permanent, measurable, not transferable or shared

User A typing the same phrase 2 users typing the same phrase

5

6

Aims

 Implement and test a

prototype of Keystroke Dynamics based on an algorithm by (Teh, Teoh et al. 2010) Signifi ificance cance

 increase security by

verifying online users who attempt to login

 reducing identity

thefts or forgery Ou Outcomes comes

 Prototype of

Keystroke Dynamics application

 Final Report and

Presentation

7

Analysis is & Docume ment ntation ation System m Integr gration ation Testi ting ng Component

• nent Test(Ve

t(Verifica ificati tion) n) Verifi fica cati tion

• n

Screen Application Component

• nent Test

t (Regi gistr tratio ation) n) Regi gistr tration ation Screen Application Setup up envi viro ronme ment Plann nnin ing

Select ect 1 Biomet metri ric: Keyst stro roke e Dyna namic mics

Select ct 2 B 2 Biome

• metri

trics cs based on ease of

implementation

• Fingerprin

ngerprint ( from Conventional & physiological group)

• Keyst

strok roke e Dynamics amics (from Emerging & behavioural group) Research search variou ious s Biometr metric ic Technologies nologies

8

• The Concept of a Feature

Vector

• Basic outline of the process
• The algorithm

9

 Dwell

ell time me (D1): ): D1= R1- P1

 Flight

ight time e (D2): ): D2 = P2 - P1

 Flight

ight time e (D3): ): D3 = P2 - R1

 Flight

ight ti time e (D4): ): D4 = R2 - R1

10

 2 parts of the process:

• Enrolment (One-time)
• Verification (Recurring)

11 Verifi rificati tion

• n Mode

Enrolmen

• lment

t Mode

 Enrolment

• Use the Feature Vector to

generate a template (mean, standard deviation, number of samples, summation)

• Store the template

 Verification

• Gaussian probability

density function (GPD) is used to compute the similarity score between the reference template and the test data template

• output score of GPD has range
• f [0,1]

12

• Basic Architectural Pattern
• Technologies Considered
• Component Design

13

 Modular design allows components to be

plugged in

14

• A1. Input/Output Layer
• A2. Application Layer
• A3. Storage Layer

High-level Architectural Pattern

• Applets
• Swings
• Java command line
• JAAS
• JavaScript

Input t / Output t Layer

• Java

Applic icatio ation n Layer

• Oracle Database

Storage age Layer

15

• Requirement:

Algorithm suggested to capture time using nanoseconds

• Out of the

technologies considered, only Swings could capture time in nanoseconds

Test t Environment

• nment

Deve velopme ment nt

16

Input t / Output ut Layer er Applica cation tion Layer er Stora rage ge Layer er

Oracle Database se Oracle Database se

• Experiment Setup
• Preliminary Results

17

 Acquisition of the test-data

• 10 volunteers from the IT department for testing
• Registration: Page to capture the user’s keystroke

patterns

• Verification: Login into the system during which the

real-time keystroke pattern was matched against the stored template to allow or deny the user.

 Captured user data useful in testing the

system

 Captured user data is also useful for

analysing patterns to obtain better algorithm

18

 With given time of 1 semester, tests

conducted with 6 volunteers were not sufficient to provide reliable metrics like FAR (False Acceptance Rate) and FRR (False Rejection Rate)

 Before a large scale testing occurs, there are

• ther issues like ethics and the user’s

consent.

 The proto

totyp type e is ready ady for more e tests ts using ing mo more e users. ers.

19

• Summary
• Lessons Learnt & Future work

20

 The Prototype of Keystroke Dynamics is

ready for further evaluation

 Components are modular to allow

replacements for other technology / algorithms / metrics

 Swings was the chosen technology as it

could capture time in nanoseconds

21

 Les

essons sons Lea earnt rnt

• technology: what other technologies could be

considered (e.g., Platform as a Service), precision of keystroke timings

• project management: how the project could have

been structured to optimize it (e.g., advertised for voluntary users in a wider population)

 Future

ture wo work

• augmenting the implementation to be more robust
• inter-operable with other key Out-Of-The-Box

(OOTB) Identity and Access Management (IAM) solutions and products

• the use of multiple biometrics

22

23

24