Communication for Omniscience.
Ni Ding, Chung Chan, Qiaoqiao Zhou, Rodney A. Kennedy, Parastoo Sadeghi · arXiv (Cornell University) · 2016
This paper considers the communication for omniscience (CO) problem: A set of users observe a discrete memoryless multiple source and want to recover the entire multiple source via noise-free broadcast communications. We study the problem of how to attain omniscience with the minimum sum-rate, the total number of communications, and determine a corresponding optimal rate vector. The results cover both asymptotic and non-asymptotic models where the transmission rates are real and integral, respectively. Based on the concepts of submodularity and Dilworth truncation, we formulate a maximization problem. The maximum is the highest Slepian-Wolf constraint over all multi-way cuts of the user set, which determines the minimum sum-rate. For solving this maximization problem and searching for an optimal rate vector, we propose a modified decomposition algorithm (MDA) and a sum-rate increment algorithm (SIA) for asymptotic and non-asymptotic models, respectively, both of which complete in polynomial time. For solving the Dilworth truncation problem as the subroutine in both algorithms, we propose a fusion method to implement the existing coordinate saturation capacity (CoordSatCap) algorithm, where the submodular function minimization (SFM) is done over a merged user set. We show by experimental results that this fusion method contributes to a reduction in computation complexity as compared to the original CoordSatCap algorithm.