Description: The main goal of this project is to carry out a systematic study of the machine learning schemes to build Intrusion Detection System (IDS)
for Industrial Control Systems (ICSs). Firstly, will be built an ICS simulation and a testbed similar to existing ICSs (SCADA, PLCs, actuators/sensors)
from the oil/gas and petrochemical industry. Second, will be built real attack scenario(s) focusing on the outcome (asset destruction, linear and/or
non-linear physical/chemical/mechanical effects with people safety and/or environmental consequences) of a sophisticated attack on a given critical
industrial process. Third, using our real attack scenario(s) and observed datasets, some of the important machine learning (ML) techniques will be
compared in the domain of IDS in terms of accuracy and efficiency. The research focuses mainly on four aspects 1) Build an ICS simulation testbed capable
to simulate real process control scenario(s) 2) Better understand the components of a sophisticated attack targeting an ICS 3) Train the ML models using observed dataset(s)
corresponding to a real attack scenario(s) to demonstrate robustness and applicability of the machine learning algorithms in real-time ICS systems, and 4) Provide recommendations
regarding IDS implementation in an ICS context as well as real-time response after detection.
2017 - present date.
Description: The main goal of this project is to develop a low-cost intelligent online parking system for public or private places.
The system will be developed using the OpenCV and JavaCV development libraries and will be optimized to run on devices that use wireless
technology such as wi-fi and cellular 3 / 4G networks.
2014 - 2015.
Description: This research project aims to develop a cross-layer scalability algorithm for uplink traffic. The scheduling algorithm should ensure QoS for real-time applications, and also the minimum resources for non-real-time applications. This algorithm takes into account the modulation used by SSs,
to use the wireless link efficiently. In addition, the scheduling algorithm interacts with a CAC mechanism that also uses a cross-layer approach
2011 - 2016.