Experimental Evaluation of Approximation Algorithms for the Minimum Cost Multiple-source Unsplittable Flow Problem

For the minimum cost multiple-source unsplittable flow problem, we propose a 2-cost and (c+ 2)-congestion approximation algorithm, where c is the number of distinct sources. We also propose some heuristics based on a linear programming relaxation with randomized rounding and a greedy approach, and implement the proposed approximation algorithms and examine the quality of approximation achieved through computational experiments.