Privacy-Preserving Data Processing Method for IoV Based on Homomorphic Conjugacy Search Problem

The Internet of Vehicles (IoV) has become a research hotspot owing to the continuous enrichment and expansion of the industrial ecology. IoV data is complex due to heterogeneity and dynamic topology, posing challenges for traditional processing methods and limited onboard device capabilities. To address this, cloud computing is essential for constructing a high-performance IoV network with accurate machine learning, extracting latent value from data. Despite cloud advances, privacy concerns in data transmission and processing within IoV persist. This paper proposed a lightweight fully homomorphic encryption algorithm to address privacy. Notably, the proposed encryption algorithm can be reduced to the conjugacy search problem (CSP) under the standard model. Based on the algorithm, a model for processing encrypted data is established and implemented on a neural network for traffic data classification. The approach is compared against conventional methods in terms of complexity, efficiency, and security. Results unequivocally demonstrate comparable accuracy with original neural networks. In contrast to traditional homomorphic encryption, the proposed approach provides equivalent security with a substantial 100-fold increase in efficiency.