Exploring Adversarial Graph Autoencoders to Manipulate Federated Learning in The Internet of Things

Mobile edge computing (MEC) enables the Internet of Things (IoT) with seamless integration of multiple application services. Federated learning is increasingly considered to improve training accuracy in MEC-IoT while circumventing the disclosure of private data, where the IoT nodes collaboratively train a machine learning model without disclosing their private data. In this paper, we propose a new cyber-epidemic attack that progressively manipulates federated learning and reduces the training accuracy of the benign MEC-IoT. The proposed cyber-epidemic attack explores adversarial graph autoencoders (GACE) to generate malicious local model updates that extract correlated features with the benign local and global models. The proposed GACE attack epidemically infects all the benign IoT nodes along with the training iterations in federated learning, while highly enhancing concealment of the attack.

I. Introduction

With the growing development of Internet of Things (IoT), mobile edge computing (MEC) is enabled to leverage powerful computing capabilities at an edge server for processing compute-intensive tasks offloaded by IoT nodes. The MEC-IoT is widely applied to a large number of applications, such as smart grids [1], intelligent transportation systems [2], and metaverse [3]. The IoT nodes upload their data to the edge server in which machine learning is used to train the IoT data. Nevertheless, this source data offloading is vulnerable to wireless attacks, such as eavesdropping [4], [5], denial of service [6], or blackhole attacks [7]. To avoid possible data privacy leakage, federated learning is studied to train a global shared model at the edge server, which aggregates local model updates instead of original training data of the IoT nodes.