Giraph-Based Distributed Algorithms for Coloring Large-Scale Graphs

The Vertex Graph Coloring problem (VGC\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathcal {VGC}$$\end{document}) is a well-known difficult combinatorial optimization problem. It is one of Karp's 21 NP-complete problems. It consists in assigning a color to each vertex of a graph in such a way that any two neighboring vertices do not share the same color, and the number of the used colors is minimized. VGC\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathcal {VGC}$$\end{document} is used to solve a variety of real-world problems such as time tabling and scheduling, radio frequency assignment, and computer register allocation. To deal with this problem on large graphs, the emerging large graph processing frameworks are an excellent promising candidate. Giraph is one of the most popular large graph processing frameworks both in industry and academia. In this work, two novel graph coloring algorithms are introduced. These algorithms designed to reap the benefit of the simple parallelization model offered by any vertex-centric frameworks, such as Giraph. The algorithms are based on well-known sequential heuristic techniques namely Largest-First (LF) and Saturation Largest-First (SLF). We have compared the performances of the proposed algorithms to previous Giraph based graph coloring algorithms, with regard to their solution quality and executing time, using benchmark graphs from the SNAP library. The obtained experimental results have revealed that the proposed algorithms are much more efficient than the existing Giraph algorithms.