Project Proposal New Project: 1/1/2013 – 5/20 ... - users at Marshall

licoricebedsΑσφάλεια

22 Φεβ 2014 (πριν από 3 χρόνια και 6 μήνες)

93 εμφανίσεις

Project Proposal


New Project:


1/1/2013



5/20/2013


Description:


This project involves
research
ing

and develo
p
ing

an open
-
source client
-
side continuous (i.e. free
-
text)
keystroke biometric authentication system

that will evolve with the user

s typing behavior
.

Keystroke
biometrics involves
measuring

an individual’
s unique typing

characteristics to provide an extra layer of
security to a computer workstation. Unlike other forms of authentication
, keystroke biometric
authentication cannot
be
cracked, lost, stolen, or given away. It does not
require extra hardware, and it is
unobtrusive to a user’s workflow.



Project Objectives:



Minimize False Acceptance Rate

(FAR)



Minimize False Rejection Rate

(FRR)



Scalability



Small CPU Footprint



Simpli
fied

User Interface



Application evolves with user


Significance/Uniqueness:


M
ost av
ailable authentication systems use static or fixed
-
text authentication whereas my project will
focus on continuous or free
-
text authentication. The systems available that provide continuous
authentication are proprietary and service is delivered through t
he cloud
(SaaS)
or on an enterprise
network server, whereas my project will develop a client
-
side system
.


My project will have multiple phases. First
,

it will extend on recent research that
provides a
novel approach to continuous keystroke authentication by using a trust threshold or confidence level.
Second
,

it will use multiple statistical distributions including Nearest Neighbor (Mahalanobis). Third
,

this project will utilize several keyst
roke biometric measurements including latency, flight time, and
dwell time. Fourth, the user templates stored in this application will evolve with the user as his or her
typing changes. However, all of this needs to be implemented in a scalable, client
-
s
ide application that
has a relatively small foot
-
print in terms of CPU and memory usage.


Approach:


I will approach this project, by first completing background research

and then by developing a template.
The temp
late will be the keystroke biometric “fingerprint”
for
each user that will be stored locally on a
SQLite database.
After that the

algorithm and distance
metrics

will be coded into the program
.
Once the
algorithm is coded, the user interface will be in
corporated into the application.

During this t
ime,

I will
test
the algorithm and tune its performance to minimize FAR and FRR.



Expected Results:

The end result of this project will be a working
Windows
application that provides continuous keystroke
authentication.
The end result will be a
scalable, client
-
side application that

evolves with the user and

has
a relatively small foot
-
print in terms of CPU and memory usage.




Budget:


$1000 Student Stipend