DAG Scheduling with Communication Delays Based on Graph Convolutional Neural Network

In vehicular edge computing (VEC), tasks and data collected by sensors on the vehicles can be offloaded to roadside units (RSUs) equipped with a set of servers through wireless transmission. These may dependent each other modeled as directed acyclic graph (DAG). The DAG scheduling problem is aimed at minimize length (makespan), i.e., maximum finish time all tasks. conventional heuristic algorithms only utilize partial information DAG, so performance these not stable. state-of-the-art method employs neural network further reduce makespan. However, this ignores fact that there are communication delays between scheduled different servers. paper, we tackle considering which makes much more challenging. Our based convolutional reinforcement learning. Experimental results show our reduces 8% 15% compared representative strategies models (GAT, GraphSAGE) 25% (HEFT, LC, CPOP) sequence-to-sequence model.