Archives

Comparative Analysis of Time Series Forecasting Models for SDMN Traffic


Anupriya and Anita Singhrova
Abstract

Major issue in the mind of administrators and engineers working in networking industry is the amount of data that should be transmitted in future on a particular network. Most challenging task is to design such a network that enhance the efficiency of network resource utilization and reserve the required resources dynamically. Dynamicity is a technique enabled by SDMN to follow variation of traffic demands in the network. In this paper, for dynamic operation of network, first traffic estimate of network bursts at different VMs is measured using time window protocol. This bursts data is captured using Wireshark. After that comparison of various time series models, Linear, polynomial, exponential and ARIMA, is preformed using same data collected through wireshark. To evaluate the forecast accuracy as well as to compare among different models fitted to a time series we have used the five performance metrics like MSE, NMSE, RMSE, MAE and MAPE are used. Based on this analysis a new model is proposed which performs better than existing models for burst data of VMs which is an approach used for SDMN in 5G networks.

Volume 11 | 09-Special Issue

Pages: 531-540

DOI: 10.5373/JARDCS/V11/20192602