ESTIMACIÓN DE VALORES DE CAMPO A PARTIR DE TRAZAS DE MEDIDAS MEDIANTE REDES NEURONALES
Fecha
2018-07-23
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Resumen
[ES]En este TFG se obtiene, mediante el entrenamiento y simulación de redes neuronales, la estimación de señal Wi-Fi en una ubicación dada, dentro de una zona geográfica delimitada. Se realiza un análisis sobre cuál sería la mejor configuración de los datos de entrada y de salida para la red neuronal, basándonos en los resultados obtenidos de promedio cuadrado de error y promedio absoluto de error (MSE y MEA). Una vez recogidos y configurados los datos para las entradas y salidas de las redes neuronales, se procede a entrenar, validar y simularlos con tres diferentes modos que ofrece Matlab. Estos son Traingd, Trainrp, Trainscg. Por último, se analizan los resultados y se realiza una comparativa acerca de que método y con qué condiciones funcionan mejor las redes neuronales; para la estimación de la red Wi-Fi y mediante una determinada ubicación.
[EN]In this TFG, the Wi-Fi signal estimation at a given location, within a defined geographical area, is carried out through the training and simulation of neural networks. An analysis was carried out on what would be the best choice of input and output data for the neural network, the results obtained from an average error and the standard error (MSE and MEA). Once the data for the inputs and outputs of the neural networks is collected and configured, an adjustment is made, validated and simulated with three modes offered by Matlab. These are Traingd, Trainrp, Trainscg. Finally, the results are analyzed and a comparison is made on which method and under what conditions neural networks work best; for the estimation of the Wi-Fi network and by means of a certain location.
[EN]In this TFG, the Wi-Fi signal estimation at a given location, within a defined geographical area, is carried out through the training and simulation of neural networks. An analysis was carried out on what would be the best choice of input and output data for the neural network, the results obtained from an average error and the standard error (MSE and MEA). Once the data for the inputs and outputs of the neural networks is collected and configured, an adjustment is made, validated and simulated with three modes offered by Matlab. These are Traingd, Trainrp, Trainscg. Finally, the results are analyzed and a comparison is made on which method and under what conditions neural networks work best; for the estimation of the Wi-Fi network and by means of a certain location.
Descripción
Palabras clave
Sistemas de Telecomunicación