Blockchain-enabled Efficient and Secure Federated Learning in IoT and Edge Computing Networks

Abstract

Federated learning (FL) has proven to be a promising solution to enable on-device machine learning over massive data generated by Internet of Things (IoT) devices at the network edge. However, the wide vulnerability space of the IoT network increases the risk of model poisoning attacks carried out by malicious or compromised IoT devices against FL model training. This paper proposes to exploit the use of blockchain technology to perform optimized monitoring of the behavior of IoT devices and select only reliable ones to provide model updates to the global FL model while preserving network performance. We formulate our worker device monitoring problem as an optimization problem and solve it to produce the optimal number of monitoring miners in the blockchain network in order to reduce the latency, bandwidth and energy consumption in the overall IoT network. Our results show that the optimal monitoring solution was able to reduce by 75% the total delay incurred by the IoT devices during training.