Decentralized Federated GAN for Hyperspectral Change Detection in Edge Computing

Abstract

Change detection on hyperspectral images is an essential task for Earth observation. Due to the vast amounts of remote sensing big data resulting from the ongoing advancements in remote sensing hardware, employing centralized learning through cloud computing emerges as a logical and convenient solution. Nevertheless, this approach overlooks the influence of the isolation and heterogeneity of remote sensing data on the reliability of change detection outcomes. In contrast, federated learning enables collaborative change detection on non-independent identical distributed remote sensing data without the need to transfer the original data. Simultaneously, it is important to acknowledge that the dependency of federated learning on the central node may pose potential data security risks. To address this issue, this paper proposes a decentralized federated learning GAN network. This ensures that raw data remains stationary during participation in unsupervised learning. Additionally, the network employs edge devices to implement federated learning, allowing adaptation to diverse devices in practical scenarios. Lastly, blockchain is integrated into a decentralized architecture for the dynamic selection of leader nodes, effectively enhancing the robustness and security of the framework. DFGAN is a novel approach for cooperative privacy preservation introduced into hyperspectral change detection systems. The method outperforms most of existing methods on three datasets, achieving more than 90% detection accuracy on all of them. Additionally, by simulating in a real on-orbit environment using a satellite constellation simulator on a Jeston TX2, FedGAN demonstrates superior accuracy in hyperspectral image change compared to existing algorithms while significantly reducing training time by 17.3%.