EdgeGuard
Graph-Augmented xAI for Threat Intelligence on Edge Infrastructure
Overview
EdgeGuard addresses the critical need for transparent, auditable threat intelligence for organizations with limited cybersecurity resources. Led by IICT-BAS in collaboration with our partner Ratio1, we are deploying a hybrid architecture that combines symbolic knowledge graphs with fine-tuned language models directly on edge infrastructure. This system transforms raw security alerts into clear, actionable recommendations that enable analysts to trace incident chains and respond effectively without relying on opaque cloud platforms.
Our motivation for participating in the ResilMesh Open Call is to validate these novel algorithms in a robust, decentralized environment. By leveraging Ratio1’s specialized edge cloud capabilities and commercial expertise, we aim to bridge the gap between advanced research and practical application, ensuring our solution effectively democratizes cyber defense while meeting the rigorous demands of critical infrastructure.
Partners
- IICT-BAS – Institute of Information and Communication Technologies, Bulgarian Academy of Sciences (Lead)
- Ratio1 – Specialized edge cloud capabilities and commercial expertise
Team
- Project Lead Prof. Dr. Tatiana Atanasova IICT-BAS
- Kalin Kopanov IICT-BAS
- Assoc. Prof. Dr. Andrei Damian Ratio1.ai
- Cristian Bleotiu Ratio1.ai
- Bogdan Ionescu Ratio1.ai
Sprints & Roadmap
S1 Design & Foundation
- G1.1 Knowledge Graph Design & Initial Population
- G1.2 Automated Data Ingestion Pipeline
- G1.3 GraphRAG Query Engine Prototype
- G1.4 ResilMesh Integration Setup
S2 Development & Integration
- G2.1 SLM Fine-Tuning
- G2.2 Explainability Framework
- G2.3 Synthetic Validation
S3 Validation & Release
- G3.1 Production Validation
- G3.2 Open-Source Release
- G3.3 Exploitation Preparation
- G3.4 Final Demonstration
Publications & Results
Details will be added soon.