Research on Secure Data Circulation Method Based on Blockchain and Privacy Computing

As the standard of privacy and security of data cross-domain circulation improves, a large number of ciphertext computation puts high demands on the communication delay and arithmetic power on the performance of the equipment, and the current privacy computation has the problems of inefficiency, federated learning needs a third party to aggregate the gradients of all parties, and it cannot avoid the discrimination of data, and so on. Therefore, this paper tries to carry out key technology research from the following three aspects: first, the architectural design of the data sharing platform based on blockchain and privacy computing is described in detail, with full consideration of versatility, and the application module adopts a layered design. Then, a multi-centered verification mechanism of federal learning model parameters is designed to achieve multi-centered verification of the training models uploaded by various institutions at the beginning of each round of global model iteration; finally, it is proposed to use Paillier homomorphic encryption in combination with parallel encryption algorithms to enhance the privacy and security of cross-domain circulation of data, and to gradually form a technical system around the safe and trustworthy circulation of data.