Research On Protection Of Industrial Sensitive Data Based On Three-layer Local/Fog/Cloud Storage

The trend of the development of manufacturing industry is intelligent manufacturing,and it is leading the global manufacturing industry reform.In recent years,there are more and more sensitive data leakage events in the industrial field.The security of industrial sensitive data has been widely studied in the world.At present,industrial enterprises lack a data security system,so a complete and effective data protection scheme is needed urgently for industrial enterprises.A protection framework for sensitive industrial data is proposed,which studies real-time industrial sensitive data and non-real-time industrial sensitive data respectively,and a threat model is defined.The main work is as follows:(1)For real-time sensitive industrial data,the sensitive data is disturbed by the improved local differential privacy algorithm M-RAPPOR,and the masked data is encoded by Reed–Solomon(RS)algorithm.And the encoded data is stored in local equipment to realize low-cost and high-efficiency data protection.The optimal solution of distributed storage in local equipment is adopted to alleviate the storage pressure on local equipment based on improving security and recoverability.According to the defined threat model of real-time sensitive data,the security analysis is conducted,which proves that the proposed scheme can provide stronger data protection for realtime sensitive industrial data.(2)For non-real-time sensitive industrial data,a cloud-fog collaborative storage scheme based on AES-RS encoding is proposed.Some encoded data was stored in the fog nodes and the rest,in the cloud nodes,to realize multilayer data protection.According to the defined threat model of non-real-time sensitive data,the security analysis is conducted,which proves that the proposed scheme can provide stronger data protection for non-real-time sensitive industrial data.Compared with traditional methods,the proposed scheme strengthens the protection of sensitive information and ensures real-time continuity of open data sharing.The location data satisfying the normal distribution is used to verify that the improved M-RAPPOR algorithm has lower computational cost.The proposed framework is evaluated via numerical experiments using location data,the turbofan engine aging dataset and Bosch's industrial dataset.The results show that the proposed scheme has better performance.