T-For: An Adaptable Forecasting Model for Throughput Performance
Ariel L. C. Portela, Silvio E. S. B. Ribeiro, Rafael A. Menezes, Thelmo P. de Araújo, Rafael L. Gomes
Abstract
Network monitoring services are performed by several companies and Internet Service Providers (ISPs), which provide results of regular performance tests, where throughput is one of the most essential information. However, the monitoring tools still need to evolve in order to encompass more complex activities, such as forecasting. Within this context, this paper presents a Throughput performance Forecasting model (called T-For), based on Neural Networks and Time Series Analysis, which estimates future network performance in specific time periods, according to past throughput measurements. The experiments, using real data from the National Education and Research Network (RNP), show that the proposed model outperformed the existing approaches, reaching high levels of forecast accuracy.