In the realm of Intelligent Railway Transportation Systems, effective multi-party collaboration is crucial due to concerns over privacy and data silos. Vertical Federated Learning (VFL) has emerged as a promising approach to facilitate such collaboration, allowing diverse entities to collectively enhance machine learning models without the need to share sensitive training data. However, existing works have highlighted VFL’s susceptibility to privacy inference attacks, where an honest but curious server could potentially reconstruct a client’s raw data from embeddings uploaded by the client. This vulnerability poses a significant threat to VFL-based intelligent railway transportation systems. In this paper, we introduce SensFL, a novel privacy-enhancing method to against privacy inference attacks in VFL. Specifically, SensFL integrates regularization of the sensitivity of embeddings to the original data into the model training process, effectively limiting the information contained in shared embeddings. By reducing the sensitivity of embeddings to the original data, SensFL can effectively resist reverse privacy attacks and prevent the reconstruction of the original data from the embeddings. Extensive experiments were conducted on four distinct datasets and three different models to demonstrate the efficacy of SensFL. Experiment results show that SensFL can effectively mitigate privacy inference attacks while maintaining the accuracy of the primary learning task. These results underscore SensFL’s potential to advance privacy protection technologies within VFL-based intelligent railway systems, addressing critical security concerns in collaborative learning environments.
Federated learning enables multiple intelligent devices to collaboratively perform machine learning tasks while preserving local data privacy. However, traditional FL architectures face challenges such as centralization and lack of effective defense mechanisms against malicious nodes, particularly in large-scale edge computing scenarios, which can lead to system instability. To address these challenges, this paper proposes a cross-zone secure federated learning method with blockchain and credibility mechanism, named CS-FL. By constructing a dual-layer blockchain network and introducing a blockchain ledger between zone servers, CS-FL establishes a decentralized trust mechanism for index detection and model aggregation. In node selection, CS-FL considers multiple dimensions, including node quality, communication resources, and historical credibility, and employs a three-stage mechanism that introduces lightweight probe tasks to assess node status before formal FL training, ensuring high-quality nodes participate. Additionally, the credibility incentive mechanism penalizes nodes that bypass probe mechanism and engage in malicious behaviors, effectively mitigating the impact of deceptive attacks. Experimental results show that CS-FL significantly improves the defense performance of FL, reducing attack success rates from 75–85% to below 5–20% when facing different types of threats, and effectively maintaining the training accuracy of the FL model. This demonstrates the potential of CS-FL to enhance the security and stability of FL systems in complex edge computing scenarios.