The results (Fig. 7) show that a higher clustering coefficient increases the potential for optimization. Surprisingly, this is not expressed by a decreasing link utilization after the optimization. Instead, the necessary latency stretch is lower for higher clustering coefficients. The behavior of the algorithm is a result of the utility function used. The utility function can therefore also be tuned according to application needs in such
case. Overall, the results show that a high clustering improves the performance of our algorithm.