A Primal-Dual based Distributed Approximation Algorithm for Prize-Collecting Steiner Tree

TL;DR

This paper introduces the first distributed approximation algorithm for the Prize-Collecting Steiner Tree problem, achieving a guaranteed approximation factor using a primal-dual approach in an asynchronous setting.

Contribution

It presents a novel distributed primal-dual approximation algorithm for PCST, adapting the centralized method to a distributed environment with guaranteed approximation bounds.

Findings

Achieves a (2 - 1/(n-1))-approximation factor.

Operates asynchronously with message complexity O(|V||E|).

First known distributed constant approximation algorithm for PCST.

Abstract

The Prize-Collecting Steiner Tree (PCST) problem is a generalization of the Steiner Tree problem that has applications in network design, content distribution networks, and many more. There are a few centralized approximation algorithms \cite{DB_MG_DS_DW_1993, GW_1995, DJ_MM_SP_2000, AA_MB_MH_2011} for solving the PCST problem. However no distributed algorithm is known that solves PCST with a guaranteed approximation factor. In this work we present an asynchronous distributed $(2 - \frac{1}{n - 1})$-approximation algorithm that constructs a PCST for a given connected undirected graph with non-negative edge weights and a non-negative prize value for each node. Our algorithm is an adaptation of the centralized algorithm proposed by Goemans and Williamson \cite{GW_1995} to the distributed setting, and is based on the primal-dual method. The message complexity of the algorithm with input graph having node set $V$ and edge set $E$ is $O(|V||E|)$. Initially each node knows only its own prize value and the weight of each incident edge. The algorithm is spontaneously initiated at a special node called the \emph{root node} and when it terminates each node knows whether it is in the PCST or not. To the best of our knowledge this is the first distributed constant approximation algorithm for PCST.