CS298 Proposal
Authentication by Mouse Movements
Shivani Hashia (shivani_hash@hotmail.com)
Advisor: Dr. Chris Pollett(pollett@cs.sjsu.edu)
Committee Members: Dr.Mark Stamp(stamp@cs.sjsu.edu) Dr.Robert
Chun (ProfessorChun@aol.com)
Abstract:
Security systems help to protect machines or sensitive data from
unauthorized users. The need for
better and cheap security systems is growing with the growth in Internet.
There are various
techniques that can authenticate a user. One of these techniques is
biometrics. In the project, we
plan to use mouse movements as the biometric for user authentication. The
mouse is used for
verifying the authenticity of the user by validating the mouse movements of
that person. There
will be two modes of identification. First will be initial login where the
user will be given a set of
points on which he has to move the mouse. Based on the pattern of mouse
movements on the
screen for those random points, he will be authenticated and allowed to
login. After the initial
login, his movements will be continuously monitored to check if he is the
specified user. Three
parameters will be used for the identification: speed of the mouse,
deviation from a straight line
and the angle of the devia tion. There will be two stages of the software
model. In the first stage,
various mouse movement patterns are recorded for each user till we find the
threshold within
which the user's mouse movement properties fall and a template file is made.
In the second stage,
actual verification will be done based on the template file created in the
first stage. The software
will compare the real time statistics with the template file and decide the
authenticity of the user.
If none of the three parameters match the user's template file data, he will
be forced to logout.
CS297 Results
- Recorded mouse coordinates for a user by making him move the mouse on
some
specified points on the screen. Plotted graph between speed and time,
deviation and time
and angle and time.
- Tracked the mouse movements in background and plotted the graph of mouse
coordinates
and time by installing mouse hooks.
- Isolated the regions on the screen where the mouse hovered the most.
Used gift-wrapping
algorithm to draw convex hulls around the dense regions.
Proposed Schedule
Week 1-3:
Aug 30-Sep 20 | Study mathematical models and implement
them to find threshhold of the parameters |
Week 4:
Sep 13-20 | Read up on Bezeir curves |
Week 5-7:
Sep 20-Oct 4 | Prepare a working model |
Week 8-9:
Oct 4-18 | Do experiments to check how well the model
works |
Week 10-12:
Oct 18-Nov 8 | Start writing the report |
Week 13:
Nov 8-15 | Submit the draft to committee |
Week 14-15:
Nov 15-29 | Prepare presentation |
Week 16:
Dec 1-8 | Oral defense |
Key Deliverables:
- Software
- Record the parameters (speed, deviation and angle) during initial login
(as
described in the abstract) by making the user move the mouse on fixed points
on
the screen and check with the already recorded parameters from previous
graphs
of these parameters that they match i.e., they fall within a certain
threshold as
calculated for that person in the first stage (described in the abstract)
using
statistical algorithms.
- For continuous monitoring of the mouse, consider the convex hulls (as
described
in CS297 Deliverable 3) as the start and end points depending on the mouse
movement at that time. Calculate the parameters (speed, deviation and angle)
between these two points and compare them with already recorded parameters
for the two points for that user.
- Do experiments to see how well the software works by gathering the data
from
various users many times and then finding the false acceptance rate and true
rejection rate of the software.
- Report
- Description of the mouse authentication model and experimental
results.
Innovations and Challenges
- Designing a user authentication technique with mouse movement has never
been used before.
- Implementing various mathematical models for finding the threshhold of
the parameters.
- Finding the false acceptance and true rejection rates.
References:
[CE93] Statistical Language Learning. Eugene Charniak. MIT. 1993.
[PI02] Circuit complexity and Neural Networks. Ian Parberry. MIT
1994
[RP03] Java-Based Internet Biometric Authentication System. Ross A.J and
Peter W. McOwan.
IEEE Transactions on Pattern and Machine Intelligence. Vol 25(9).Page
1166-1172. 2003.
[HR02] Movement Awareness for Ubiquitous Game Control. Robert Headon &
Rupert Curwen.
Personal and Ubiquitious Computing. Springer-Verlag. Vol 6. Page 407-415.
2002.
[RS02] Artificial Intelligence.Stuart J.Russell , Peter Norvig. Prentice
Hall. 2002
|