Secure Federated Graph Learning With Traffic Throttling and Flow Scheduling

Federated learning has been proposed as a promising distributed machine learning paradigm with strong privacy protection on training data. In this paper, we study federated graph learning (FGL) under a cross-silo setting where several servers are connected by a wide-area network and cooperate in training a graph convolutional network (GCN) model. We find that communication becomes the main system bottleneck because of frequent information exchanges among federated severs and limited network bandwidth. To tackle this challenge, we design S-Glint, a secure federated graph learning system, containing a homomorphic encryption-based protocol with pre-aggregation and batching strategies. The network traffic throttling and flows scheduling are also proposed to accelerate the learning process. To evaluate the effectiveness of S-Glint, we conduct experiments using trace-driven simulations. The results show that S-Glint can significantly outperform existing federated learning solutions.