Forecasting Internet Demand Using Public Data: A Case Study in Brazil

Abstract

In Brazil, the government has historically given low attention to the planning of telecommunication infrastructure, such as the prediction of the Internet bandwidth in the short and medium term, since this process can be slow and costly. Notably, smart city applications are impaired by this policy, because they depend on cost-benefit technology to support the Internet of Things. This paper presents a method for forecasting the Internet demand based on public data obtained from the International Telecommunication Union, the World Bank, and government agencies, using Brazil as a case study. The information inputs are associated with the population growth, and with social and technological development. The prediction process uses statistic concepts and models to infer the relationship between input variables and the data bandwidth rate. The proposed methodology is not restricted to the prediction of the Internet demand and may also be used to estimate other concerns for developing countries, such as oil, energy, and water consumption. The method is compared with other time series analysis models. The results reveal that factors related to innovation and technology significantly impact the annual projection of the Internet demand.

Keywords

• Smart cities
• telecom infrastructure
• dynamic regression
• Internet demand forecasting