Privacy-Preserving Intelligent Financial Information Processing in Cloud-Edge Distributed Systems Based on Homomorphic Encryption and MapReduce

Privacy-preserving financial big-data processing must simultaneously control plaintext exposure and the computational cost of distributed analytics. This paper proposes a cloud-edge MapReduce-ECC framework in which edge terminals perform local validation, sharding, feature fusion, and ciphertext generation, while cloud Reduce nodes aggregate only matched protected shards and HDFS stores encrypted blocks with rack-aware replication. The design clarifies key separation, ciphertext matching, fault recovery, and compliance-oriented data-localization rules. Experiments on a 5.5-TB operational e-commerce financial dataset show that 42 Map nodes provide the best observed parallel-processing trade-off. The HDFS deployment maintains storage-task completion and post-storage availability close to 100%, while the stated trust-boundary design prevents plaintext financial shards from being transmitted to ordinary cloud workers. These results indicate that the framework provides a scalable privacy-oriented processing pipeline for distributed financial information systems.