A Secure Verifiable Computation Scheme for Matrix Determinants Based on Dual Perturbation Encryption and Multi-server Parallel Computation
Tianpeng Zhang, Zhiyu Ren, Qiuli Wang · 2023
With the development of cloud computing and IoT technologies, there is an increasing demand for timeliness in secure verifiable computing schemes. On one hand, the computation time can be reduced by improving the efficiency of client data encryption and decryption. On the other hand, the computation time of servers can be reduced by leveraging the advantages of multi-server parallel computing. The secure verifiable computing scheme for matrix determinant based on dual perturbation encryption and multi-server parallel computing achieves both reduced computation time for client matrix encryption and decryption, as well as decreased computation time for servers. In this scheme, a dual perturbation encryption algorithm is proposed based on specially constructed upper or lower triangular sparse matrices, which improves the efficiency of matrix encryption and decryption computation while meeting the same security goals. Additionally, a multi-server parallel computing algorithm is introduced based on matrix partitioning and matrix LU factorization techniques, which extends the available server quantity from two to multiple and enhances server parallel computing efficiency. Performance analysis indicates a 50% improvement in matrix encryption and decryption computation efficiency, and a 30% increase in parallel computing efficiency as the number of servers increases.