Secure Coded Multi-Party Computation for Massive Matrix Operations

In this article, we consider a secure multi-party computation problem (MPC), where the goal is to offload of an arbitrary polynomial function some massive private matrices (inputs) cluster workers. The workers are not reliable. Some them may collude gain information about input data (semi-honest workers). system initialized by sharing (randomized) each matrix server. Since massive, share's size assumed be at most 1/k fraction matrix, for k ∈ \mathbb N. objective minimize number needed perform task correctly, such that even if subset t-1 workers, t N, collude, they cannot any matrices. We propose scheme, called sharing, and show it admits basic operations as adding multiplication transposing matrix. By concatenating procedures operations, can calculated, subject constraints. proposed scheme offer order-wise in terms needed, compared approaches formed concatenation job splitting conventional MPC approaches.