Panacea

Research Units: University of Rome 3 University of Rome La Sapienza University of Modena and Reggio Emilia

Duration September 2023 - September 2025

Research Goals:

1. Create an explainable anomaly-based IDS for a large-scale computer system, where each element (hardware or software component):
   • Has an autonomous behavior described by a multivariate time series with different sampling rates.
   • Operates under possibly unknown conditions.
2. Develop a robust, scalable, and multi-agent Intrusion Response System that:
   • Is based on a joint model of defender and attacker.
   • Ensures robustness with respect to model changes.
3. Establish the first publicly available testbed for validating Self-Protecting Systems.

Research Contributions:

• The first integrated approach to self-protection, combining explainable anomaly-based intrusion detection and intrusion response.
• Utilization of neural network models for local intrusion detection and post-hoc models for explanations.
• Implementation of a stochastic game with a Generative Adversarial Network for scalable and adaptive intrusion response.
• Creation of the Panacea framework for realizing self-protecting systems, validated on a large-scale web-based system.

Impact and Dissemination:

• Advances in cyber-security automation, particularly in Anomaly Detection, Intrusion Response, and Secure Software and Platform Engineering.
• Dissemination of scientific results to the research community.
• Development of new course material for undergraduate and graduate curricula in cyber-security, artificial intelligence, and big data.

Future Applications:

• Techniques applicable to other domains, including Cloud-Edge continuum systems, critical infrastructures, and Cyber-Physical Systems.