Revisiting Privacy-Preserving Min and k-th Min Protocols for Mobile Sensing

Exploiting participants’ data without knowing them is the center topic of secure mobile sensing data aggregation. In this article, we study how to improve existing protocols for computing the minimum or
$k$
-th minimum of all participants’ data in a privacy-preserving manner. Existing protocols for these two computations require relatively high communication cost or frequent interactions, which leads to unbearable time consumption when the network delay is high. We improve the min computation protocol proposed by Zhang et al. 2017, cutting down on its need for interactions and thus making it perform better in terms of efficiency. We also propose a new protocol as a secure substitute of the Bit-choosing Algorithm designed by Yu et al. 2018. It helps participants generate a secret permutation, which will be further used in the
$k$
-th min computation protocol to achieve higher accuracy and efficiency. Theoretical analyses are done to help predict and understand our new protocols’ performances, and later evaluations show that both these new protocols perform notably better in comparison to the existing ones.