No abstract

Cloud storage is an essential method for data storage. Verifying the integrity of data in the cloud is critical for the client. Traditional cloud storage approaches rely on third-party auditors (TPAs) to accomplish auditing tasks. However, third-party auditors are often not trusted. To eliminate over-reliance on third-party auditors, this paper designs a blockchain-based auditing scheme that uses blockchain instead of third-party auditors to ensure the reliability of data auditing. Meanwhile, our scheme is based on the audit method of the quad Merkle hash tree, using the root of the quad Merkle hash tree to verify the integrity of data, which significantly improves computing and storage efficiency. Automated verification of auditing activities by deploying smart contracts on the blockchain allows us to have a more up-to-date picture of data integrity. The performance of the scheme is evaluated through security analysis and experiments, which prove that the proposed scheme is secure and effective.INDEX TERMS Integrity auditing, Blockchain, Merkle tree, Smart contract.

Increasing numbers of users are outsourcing data to the cloud, but data integrity is an important issue. Due to the decentralization and immutability of blockchain, more and more researchers tend to use blockchain to replace third-party auditors. This paper proposes a data integrity system based on blockchain expansion technology that aims to solve the problem of high cost for blockchain network maintenance and for user creation of new blocks caused by the rapid growth of blocks in the data integrity audit scheme of existing blockchain technology. Users and cloud service providers (CSP) deploy smart contracts on the main chain and sub-chains. Intensive and frequent computing work is transferred to the sub-chain for completion, and the computation results of the sub-chain are submitted to the main chain periodically or when needed to ensure its finality. The concept of non-interactive audit is introduced to avoid affecting user experience due to the communication with the CSP during the audit process. In order to ensure data security, a reward pool mechanism is introduced. Comprehensive analysis from aspects such as storage, batch auditing and data consistency proves the correctness of the scheme. Experiments on the Ethereum blockchain platform demonstrate that this scheme can effectively reduce storage and computational overhead.INDEX TERMS Blockchain, cloud storage, data auditing, blockchain expansion.

In cloud storage mode, users lose physical control over their data. To enhance the security of outsourced data, it is vital to audit the data integrity of the data owners. However, most of the current audit protocols have a single application scenario and cannot accommodate the actual needs of individuals and enterprises. In this research, a safe and efficient auditing scheme is proposed that is based on a hierarchical Merkle tree. On the one hand, we use a hierarchical authentication data structure and local signature aggregation technique to reduce the scale of the Merkle tree. In addition, authoritative nodes are introduced to reduce the length of the authentication path and improve the update efficiency. On the other hand, we introduce a monitoring mechanism that is based on the original data integrity auditing model to analyze the cloud data, which improves the transparency and credibility of cloud service providers. In addition, we achieve incomplete data recovery through log analysis, which greatly reduces the number of replicas of files under the premise of multi-copy auditing, reduces the burden on cloud service providers, and improves the fairness of audit protocols. The theoretical analysis and experimental comparison prove that the method is secure and efficient. It can effectively reduce the computational overhead and storage overhead in integrity auditing.

Distributed federated learning models are vulnerable to membership inference attacks (MIA) because they remember information about their training data. Through a comprehensive privacy analysis of distributed federated learning models, we design an attack model based on generative adversarial networks (GAN) and member inference attacks (MIA). Malicious participants (attackers) utilize the attack model to successfully reconstruct training sets of other regular participants without any negative impact on the global model. To solve this problem, we apply the differential privacy method to the training process of the model, which effectively reduces the accuracy of member inference attacks by clipping the gradient and adding noise to it. In addition, we manage the participants hierarchically through the method of trust domain division to alleviate the performance degradation of the model caused by differential privacy processing. Experimental results show that in distributed federated learning, our designed scheme can effectively defend against member inference attacks in white-box scenarios and maintain the usability of the global model, realizing an effective trade-off between privacy and usability.