Most of network management tasks in traffic engineering such as traffic scheduling, path planning, both of them are required the accurate and fine-grained network traffic. However, it is difficult to capture and estimate the volume of network traffic due to its time-varying nature. In this paper, we study the network traffic estimation scheme to estimate the fine-grained network traffic. Firstly, the network traffic is constructed as a time series and the autoregressive moving average (ARMA) method is used to characterize and model network traffic. Secondly, in order to decrease the estimation errors of the ARMA model, we use the optimization theory to adjust the estimation results. We construct an objective function with constraints. We find that objective function is an NP-hard problem, then we introduce a heuristic algorithm to find the optimization results. Finally, to evaluate the performance of our proposed scheme, we construct a simulation platform and compare our scheme with that of the other methods in an SDN simulation platform. The simulation results indicate that our approach is effective and our method can reflect the network traffic characteristics.
The different versions of the original document can be found in: