Dynamic Graph Neural Networks for Real-Time Failure Recovery

This research introduces the Dynamic Graph Neural Network for Real-Time Failure Recovery (DGNN-RTFR) framework, a novel approach to network failure recovery specifically designed for Software-Defined Networking (SDN) environments supporting Industrial Internet of Things (IIoT) applications. Traditional failure recovery mechanisms suffer from slow convergence times, limited multi-failure handling capabilities, and inadequate optimization for industrial requirements, making them unsuitable for mission-critical industrial environments.

Objectives

1. Develop a unified spatio-temporal graph learning model that simultaneously captures network topology and temporal evolution.
2. Design and implement an advanced multi-failure recovery algorithm capable of handling multiple concurrent failures.
3. Create a robust prediction mechanism that maintain accuracy under noisy and incomplete telemetry conditions.
4. Formulate IIoT-specific optimization criteria that ensure compliance with industrial constraints.
5. Deliver a computationally efficient solution suitable for deployment in fog computing environments with strict processing constraints.

Publications

• (Paper) HIFFR: Hybrid Intelligent Fast Failure Recovery Framework for Enhanced Resilience in Software Defined Networks
• (Paper) DYNAPARC: AI-Driven Predictive Path Failure Management for Industrial IoT-Fog Networks