Overview

Research Topics

Research Projects

Infrastructure

Scientific Contributions

Research Prototypes

 Software Systems

Research Prototypes

SMART
SMART is a Linux variation that provides a platform for:

• real-time detection of application exploits that inject code into the address space of running processes (stack/heap overflows, etc.);
• detection and documentation of 0-day exploits in honeypot systems;

SMART aims at enhancing the dependability of applications that run on top of Linux systems. It comprises a userland toolchain and a syscall monitoring and policy enforcement subsystem that is incorporated into the Linux kernel. SMART can be used as a secure computing platform for guarding applications running on top of it against malicious attacks, making it thus an ideal host for servers running critical services.

Software Configuration Manager
SCM is a platform for the automatic configuration of in-home networked devices. It is capable of automatically configuring in-home devices by making use of ontology bases that contain device attributes and properties. The prototype constructs configuration scripts that are executed within the home environment for appropriately configuring in-home devices.

Mobile Porter
The Mobile Porter prototype comprises a methodology, a platform, and an accompanying toolset for the efficient automatic porting of mobile applications and games to a range of mobile devices.

WSN Policy Formulator and Enforcer
The WSN PFE allows the dynamic formulation and enforcement of business level policies on data collected from building spaces through a deployed WSN. The intelligent processing of the data allows the drawing of intelligent conclusions and the selective enforcement of actions as specified in the formulated policies.

GetTogether
G2G is a proximity-awareness platform for Mobile Information Device (MID) users that supports various applications in mobile social networking, personalized recommender systems and gaming. User proximity to another friend-user or a place is determined via GSM cell information. Overall, the system allows users to see nearby friends, or to leave notes and/or recommendations valid only in their current location. The system’s flexibility allows for users to set-up their own treasure-hunt games and other applications.

Focused Crawler
This system allows a user to submit a set of "topical" URLs containing documents related to the topic of the focused crawling (training instances), and then, the system, starting from a set of "seed" URLs starts a crawling throughout the web, looking for relevant pages with respect to the topics discussed in the initial set of "topical" URLs.

CPACE
Continuous Profile Adaptation through Classifier Ensembles. CPACE is a system that continuously monitors TV-viewer choices, and using machine learning techniques, it classifies not-seen-before content from an Electronic Program Guide (EPG) and suggests to the user content that he/she is likely to be interested in viewing. Unlike collaborative filtering approaches that suffer from privacy related issues, our architecture offers enhanced privacy by not requiring viewers’ data to be available or stored anywhere except the viewer’s Set-Top Box. The system exhibits very high precision in its recommendations in a real-world installation where it usually achieves rates of more than 90%, and for a significant fraction of the user population it achieves 100% correct recommendations.