Coded Distributed Computing Schemes via Grouping Method
Jing Jiang, Lingling Zhou, Lingxiao Qu · 2022
Coded distributed computing introduced by Li et al. in 2015 is an efficient approach to reduce the communication load in general distributed computing frameworks such as MapReduce. But there is a big disadvantage in this scheme, that is, the numbers of required input files and output functions increase exponentially with the increase of the number of computing nodes. In this paper, we make an attempt to overcome these problems by constructing some new coded distributed computing schemes. We demonstrate that when the number of computing nodes becomes large, 1) the number of input files in the proposed scheme is much smaller than that of input files in the scheme derive by Li et al.; 2) the required number of output functions is reduced exponentially. Meanwhile, the ratio of the communication load to that of the scheme derived by Li et al. is less than 3.