Baobao Chai

Federated consensus eliminates traditional resource waste by leveraging participant nodes for collaborative model training. However, conventional schemes typically allocate block creation authority based on node performance, inevitably centralizing block generation power. Crucially, in Non-independent and Identically Distributed (Non-IID) environments, data heterogeneity empowers a fraction of feature-rich nodes to dominate the iterative updates. Furthermore, during the training phase, the global model is highly susceptible to trapping in skewed local optima, ultimately securing high accuracy on a handful of clients at the implicit expense of others. To address these challenges, we propose LayerPert, a fair federated consensus scheme based on layer-wise gradient perturbation, which prioritizes mitigating client disparities while sustaining competitive global model performance. Specifically, we design a layer-wise gradient perturbation mechanism that injects random noise into model parameters, allowing the model to escape local optima and improve generalization. In addition, a multi-objective optimization framework is employed to jointly improve global prediction accuracy and ensure fairness performance across clients. We further provide a rigorous convergence analysis of LayerPert. Extensive experiments on two public datasets demonstrate that LayerPert consistently outperforms existing fair federated learning schemes in both predictive performance and fairness.

In the current hypergrowth phase of the Internet of Things, cross-domain data access becomes more and more frequently. Whereas, the lack of trust between domains makes cross-domain data access extremely hard. Traditional schemes typically depend on a third party to establish trust between data accessing entities, which can easily result in single point of failure. To conquer the aforementioned challenge, this paper proposes BSCDA, a blockchain-based cross-domain data access scheme designed to enable secure data transmission across domains. The decentralization, transparency, and anti-tampering features of blockchain perfectly solve the issue of single point of failure and foster trust among various domains. Specifically, a certificate management method is developed to address the certificate storage issue by leveraging a mapping table to store the revocation certificate index on the blockchain. This method not only ensures the verifiability of the certificate but also reduces the storage overhead. Additionally, a four-party key agreement mechanism is designed to guarantee the secure data transmission during the process of cross-domain data access. Security analysis prove the feasibility of our proposed scheme. Extensive experiments demonstrate the superiority of our scheme in cross-domain data access.

Public clouds have drawn increasing attention from academia and industry due to their high computational and storage performance. Attribute-based encryption (ABE) is the most promising technology to simultaneously achieve confidentiality and fine-grained access control of the cloud-stored data. However, traditional ABE that relies on centralized authority faces several key management issues, such as the key escrow, key distribution, key tracking, key update, and heavy communication and computing overhead for users, which will cause security concerns and impede its widespread application. On the other hand, blockchain technology preserves distributed ledgers to ensure the immutability and transparency of data, which can further solve the security vulnerabilities caused by system centralization. This paper proposes a blockchain-assisted transformation method to solve all the key management problems mentioned above in ciphertext-policy ABE by utilizing technologies such as secret sharing protocols. In addition, our transformation method realizes two additional benefits: outsourced decryption and efficient user revocation, which are extremely valuable for practical implementations. We simulate a demonstration by adopting the most popular permissioned blockchain, Hyperledger Fabric. The security and efficiency analysis reveals that the scheme obtained from our transformation method can achieve replayable chosen-ciphertext security with extremely efficient decryption.

With the rapid development of cloud computing, a large number of web services have been emerging quickly, which brings a heavy burden for users to choose the services they preferred. In order to suggest web services for users, recommendation algorithms are needed and many of them have been investigated recently. However, most of the existing recommendation schemes are based on centralized historical data, which may lead to single point of failure. Generally, the data contains a lot of sensitive information that cloud may expose the privacy of users, which makes most cloud platforms reluctant to share their own data. In order to solve the above issues, the secure data sharing among cloud platforms is necessary for better recommendation, which can maximize the profits. In this paper, we propose a blockchain-assisted collaborative service recommendation scheme ( $BC - SRDS$ ). Specifically, we adopt the ciphertext-policy attribute-based encryption (CP-ABE) algorithm to encrypt the data, which ensures the data confidentiality and realizes secure data sharing. Then, we utilize the blockchain to share data, such that the DoS attack, DDoS attack and single point of failure can be avoided. Meanwhile, the data integrity, tampering-proof of data are guaranteed through the blockchain. And we use locality-sensitive hashing algorithm to recommend the services for users. Finally, it is proved through the security analysis that $BC - SRDS$ is capable of achieving data confidentiality, data integrity and tampering-proof. A series of experiments show that $BC - SRDS$ achieves better recommendation accuracy compared with the existing schemes.