Secure two-party queries with efficient sorting
Chuanwen Lin, Dikang Han, Tianlun Meng · 2025
Currently with the development of the big data industry, collaborative data analysis is fundamental. However, collaborative data analytics faces privacy protection issues. Different data holders must interact data with each other, leading to privacy problems. Now, secure SQL based on Secure Multi-Party computation(MPC) can protect the privacy of input data. However, the complexity of secure SQL with sort is usually 0(n log n)or 0(n log2 n). To improve the performance of secure SQL, this paper focuses on the efficient implementation of secure sort. We implement bitonic sort and quicksort in mpc4j. The experiment results show that when dealing with large-scale datasets, bitonic sort with bitonic inputs is roughly 4.3 times more efficient than quicksort. Whereas, if the inputs are unordered, the quicksort is more efficient than the bitonic sort.