GNN Link Prediction for Time-Triggered Systems

Abstract

Research on graph neural networks (GNNs) has increasingly gained popularity recently. GNN is considered a powerful tool for solving machine learning tasks that require dealing with irregular topologies such as graph data. Meanwhile, solving the scheduling problems for time-triggered systems has been debated for a long time. Even though several algorithms were proposed to solve this problem, none considered exploiting GNN partially or wholly, solving time-triggered scheduling. In this work, we propose an approach for dynamic adaptation in time-triggered systems using GNN. We use GNNs to solve scheduling problems for time-triggered systems by transforming job allocation probelms to link prediction tasks. The preliminary results show that GNNs have a promising potential to perform job allocation problems in time-triggered systems.