Multimodal data privacy protection and completeness verification method for mobile crowd sensing

Most privacy-preserving approaches for mobile crowd sensing systems consider the privacy of single-modal data, while the data that sensing equipment may be sensing is frequently multimodal. Therefore, this paper proposes a multimodal data privacy protection and completeness verification method for mobile crowd sensing. Firstly, a cross-attention mechanism is utilized to fusion multimodal data to facilitate access to data information across various modes, improving the accuracy and reliability of data encryption. Secondly, a scalable superincreasing sequence is applied to store the multimodal data gathered by each sensing user, and the multimodal data is encrypted using an upgraded Paillier algorithm to prevent malicious attackers from obtaining the data information. Then, each ciphertext provides a validator using the Boneh-Lynn-Shacham signature algorithm. The sensing platform can apply a verification code to validate the completeness of the aggregated ciphertext data to ensure that the encrypted multimodal data has not been changed before being decrypted. Finally, experimental results demonstrate that the method proposed in this article not only effectively protects the privacy of multimodal data but also minimizes Communication costs and computational overhead.