Current Research Projects

Project: Leveraging Generative Adversarial Networks for Enhanced Cybersecurity in Smart Transportation Systems

Investigators: Dr. Mohamed Rahouti (Fordham University), Prof. Moussa Ayyash (Chicago State University), and Mr. Ali Alfatemi (Fordham University)

Summary: The purpose of this collaborative project is to forge a transportation ecosystem that is not only intelligent and efficient but also inherently secure against an array of cyber threats. This project will utilize the capabilities of Generative Adversarial Networks (GANs) along with other generative AI methodologies to proactively identify, simulate, and neutralize cyber threats against sophisticated transportation networks. The project combines the strengths of GANs, Reinforcement Learning (RL), and differential privacy to significantly devise a cybersecurity framework for resilient and advanced transportation systems.

This project does not only align with CARNATIONS's mission of promoting secure and resilient navigation in advanced transportation systems but also sets a new benchmark for cybersecurity practices across the industry.

Technical Framework:

1. Implementation of threat simulation and detection using GANs: By generating synthetic attack scenarios that closely resemble a variety of cyber threats, such as GPS spoofing and sensor tampering, these networks will play a crucial role in uncovering potential vulnerabilities within the transportation infrastructure. Additionally, this simulated data will be instrumental in training our detection algorithms to recognize and respond to subtle indicators of cyber-attacks with high precision,
2. Development of adaptive defense mechanisms powered by RL: RL will help in creating defense algorithms that are capable of learning and evolving based on both simulated attacks and real-world data. The goal is to establish a dynamic and responsive defense system that can adjust its strategies to counter new and evolving cyber threats, thereby ensuring the continuous protection of navigation and control systems within smart transportation networks.
3. Integration of robustness testing through the application of differential privacy techniques during the model training phase: This approach is designed to prevent our generative models and defensive algorithms from overfitting to specific attack scenarios or inadvertently leaking sensitive data. Incorporating differential privacy is expected to significantly enhance the resilience of our cybersecurity measures against data manipulation attacks and other sophisticated threats.