ABSTRACT Federated Learning allows multiple clients to train local models and aggregate them on the server side. The client is invisible to the shared global model generated by the server, which provides an opportunity for malicious attackers to utilize the inherent vulnerability of federated learning to initiate data leakage attacks. Existing attack techniques are largely client‐based and focus on inferring model parameters directly, but do not work for server‐based attacks, mainly due to differences in their ability to generalize attacks. Yet few robust data leakage attacks toward federated learning vulnerability have been developed on the server side. To address the above problem, we propose MOFDRNet, a Multi‐Objective Fake Data Regression Network model that integrates the loss function and multiple metrics strategies. The key idea is to deploy a malicious attack model on the server with the purpose of generating fake data and labels and continuously approximating the shared gradients between clients and the server, thereby recovering clients' private data. Experimental results demonstrate that the MOFDRNet model has significant advantages in implementing data leakage attacks. Finally, we also discuss the differential privacy defense approach in this study.

Federated Learning (FL), a secure and emerging distributed learning paradigm, has garnered significant interest in the Internet of Things (IoT) domain. However, it remains vulnerable to adversaries who may compromise privacy and integrity. Previous studies on privacy-preserving FL (PPFL) have demonstrated limitations in client model personalization and resistance to poisoning attacks, including Byzantine and backdoor attacks. In response, we propose a novel PPFL framework, FedRectify, that employs a personalized dual-layer approach through the deployment of Trusted Execution Environments and an interactive training strategy. This strategy facilitates the learning of personalized client features via private and shared layers. Furthermore, to improve model’s robustness to poisoning attacks, we introduce a novel aggregation method that employs clustering to filter out outlier model parameters and robust regression to assess the confidence of cluster members, thereby rectifying poisoned parameters. We theoretically prove the convergence of FedRectify and empirically validate its performance through extensive experiments. The results demonstrate that FedRectify converges 1.47-2.63 times faster than state-of-the-art methods when countering Byzantine attacks. Moreover, it can rapidly reduce the attack success rate to a low level between 10% and 40% in subsequent rounds when confronting bursty backdoor attacks.