In my current role within the CARE
center I supervise many projects. The following list are some of my
current and past projects.
CyberSecurity Testing Protocol - We
developed a protocol and sample implementation for a CyberSecurity
Defense Evaluation system. The system will facilitate quick and
accurate testing of new defense software through testing new
vulnerabilities created by our team.
Behavioral Biometrics-Based User Authentication - Measuring user behaviors to determine if the user is the one currently authorized on the Android device. (Funded by DARPA)
P2P Network for Forensics Investigations
- Building a P2P network fabric to support large-scale network forensic
queries. This will provide a drastic speed-up during investigations by
providing investigators information in minutes versus days/weeks. (PI
funded by DARPA)
Malware Analysis - applying
crowdsource data to analyze unknown malware binaries. Using open source
data to automatically create capability mappings through static
analysis. (PI funded by DARPA)
Cloud-based DDoS Defense -
implementing a moving target-based DDoS defense system in the cloud.
This system is able to resist large scale DDoS attacks with fewer
system resources than static overlay approaches. (Funded by DARPA)
Anomaly-based Intrusion Detection
- a Network Intrusion Detection System (NIDS) using novel
anomaly detection. Currently achieved better TPR/FPR than other
approaches.
Resisting resource exhaustion -
implemented a system to dynamically monitor binaries to detect and
reverse malware code injection attacks on the usually benign programs.
This tool is built on top of Intel's PIN tool and Stanford's KLEE
symbolic engine. (Funded by IARPA)
My previous research in computer vision focussed on feature matching
for wide baseline image
pairs. Using a combination of current approaches from literature and
our own algorithms developed at GMU we worked to enhance
feature matching in the context of rigid man-made structures. A typical
usage we solved is two different people take a picture of
a building from different viewpoints at different times and determining
they are of the same building. Additionally, can you determine in a
large corpus of building pictures which are the same and which are not?
I have also worked on different
vision-based tracking techniques for a
variety of scenarios. Tracking vehicles from UAV imagery to tracking
human motion to create Human-Computer interfaces using optical sensors
are all interesting problems I am working to solve.