Forecasting key performance indicator of mobile networks: application to mobile cellular networks

• Autor(a) principal: Sousa, Ângelo Miguel Raposo Almeida e
• Data de Publicação: 2019
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/29563
• topic: Mobile networks; Forecasting; Exponential Smoothing; KPI; Time Series; ARIMA; Random-Walk

The increase of data trafic in the world has increased the need for mobile network operators to take greater care in planning and managing theirs infrastructures. This work explores the performance of several statistical forecasting models aplied in voice and data trafic. This data was obtained from an European mobile network, and, regarding the predictive models, it was applied classic models like exponential smoothing, Holt-Winters, ARIMA, Random-Walk; as well as two more recent model proposals. Regarding the daily data, the proposed model could predict values with higher precision compared to the other models. For hourly data, depending on the time zone where the models were tested, the models with higher performance were Random-Walk and the second proposed model. In summary, this dissertation shows the performance of several classic statistical models, and how these compare to recently proposed models. It also shows that mobile network operator can use statistical forecasting methods to try to get information on how their network might react in future, giving valueable insights to perform a better managment of their network.