Keyword Searchable Provable Data Possession in Generative AI Enabled Intelligent Transportation Systems

In generative artificial intelligence (AI) enabled intelligent transportation systems (ITS), real-time interaction and analysis of massive multi-source data are the core drivers for system optimization. However, large scale outsourcing storage of data, while enhancing computational efficiency, also poses security threats such as data integrity compromise, sensitive information leakage, and malicious tampering. These issues severely impact the quality of generative AI model training and the reliability of system services. Although existing cloud provable data possession (PDP) schemes can verify data integrity, they incur high computational overhead in ITS scenarios and lack support for fine-grained data retrieval. To address the need for specific data auditing by data users in generative AI enabled ITS, we propose a keyword searchable provable data possession scheme, which integrates data integrity verification with efficient data retrieval. First, a lightweight indexing approach is integrated to optimize the keyword search process, allowing different data users to initiate keyword based targeted queries based on their specific business needs, with integrity verification performed only on matching data blocks. Second, an efficient integrity auditing scheme is designed to ensure the integrity of outsourced intelligent vehicle data. Finally, security analysis and experimental results demonstrate that the proposed scheme achieves acceptable efficiency in computation overhead compared to existing works, while improving the auditability and searchability of outsourced ITS data.