Abstract

As massive real-time systems and various types of networks connect to the Internet, real-time Internet connections are becoming increasingly prominent. Research on bottleneck delays provides a reference for solving real-time performance issues in data transmission for real-time systems on the Internet. In this study, under the Internet macro-topology, we utilized real probing data to conduct an in-depth study on the evolution behavior of bottleneck delays in end-to-end delays. We demonstrate that more than 90% of valid paths were 9 hops on average, and the ratio of the maximum link delay to the one-way delay was above 1/3. In addition, bottleneck delays on the valid paths were mainly concentrated in the range of [0, 130 ms], and the relative spatial positions of the bottleneck delays appearing in the links of one-way delays tended to be near the middle. The timing evolution sequence of the bottleneck delays exhibited a slow upward trend and quasiperiodic oscillations. On this basis, the Logistic equation was introduced to describe this evolution and establish the prediction model, and the gray Wolf Optimizer (GWO) algorithm was used to fit and optimize the model parameters. Finally, experiments were conducted to verify the validity and accuracy of the models. Their prediction accuracies exceeded 90% according to a quantitative evaluation, and the models fit well according to a χ2 goodness-of-fit test. This validates our modeling approach and demonstrates that the model can predict bottleneck delays accurately in a short period of time.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call