Please use this identifier to cite or link to this item:

http://hdl.handle.net/10609/87285
Title: Gestión de bandas de frecuencias en entornos celulares mediante técnicas predictivas de deep learning
Author: Parra Guirado, Andrés
Director: López Vicario, José
Tutor: Vilajosana i Guillén, Xavier
Keywords: LSTM
deep learning
TensorFlow
Issue Date: Jan-2019
Publisher: Universitat Oberta de Catalunya (UOC)
Abstract: The purpose of this work is to implement a software capable of predicting calls made in cellular networks in the city of Milan. The sources of traffic data are provided by the city's Italia Telecom generated by its users as well as by users displaced to this area. Implementation is possible with different Deep Learning techniques and the TensorFlow framework and, to achieve this with sufficent prediction horizon, use of an LSTM network is advisable, as these networks have large long-term Memory. In the tests performed, the LSTM network was compared with a multilayer perceptron. As expected, LSTM network showed the best performance. Based on the product developed in this work, the activation/deactivation of frequency bands in Frequency Shift Repeaters (FSR) could be improved when these are going to carry a higher number of calls than usual. As a result, we have been able to make predictions with a prediction horizon of 18 samples, which with the time interval of the data used is equivalent to three hours of prediction. This is a sufficient horizon to satisfy the proposed application.
Language: Spanish
URI: http://hdl.handle.net/10609/87285
Appears in Collections:Bachelor thesis, research projects, etc.

Share:
Export:
Files in This Item:
File Description SizeFormat 
torettoTFM0119memoria.pdfMemoria del TFM4.39 MBAdobe PDFView/Open
torettoTFM0119presentación.pdfPresentación del TFM1.37 MBAdobe PDFView/Open

Items in repository are protected by copyright, with all rights reserved, unless otherwise indicated.