A study on Improving Latency-Optimized Fair Queuing Algorithm

WFQ (Weighted Fair Queuing) is the most popular fair queuing algorithm, but it had the inherent drawback of a poet bandwidth utilization, particularly under the traffic requiring a low rate but tight delay bound such as internet phone. It was recently identified that the poor utilization is mainly due to the non-optimized latency of a flow and then LOFQ(Latency-Optimized Fair Queuing) to overcome the drawback was introduced. In this paper, we improve the performance of LOFQ by introducing an occupied resource optimization function and reduce the implementation complexity of recursive resource transformation by revising the transformation scheme. We also prove the superiority of LOFQ over WFQ in terms of utilization. The simulation result shows that the improved LOFQ provides higher utilization than that in the legacy LOFQ.