Análisis de Tráfico en Dispositivos Móviles mediante Técnicas de Aprendizaje Profundo Supervisado.
Loading...
Official URL
Full text at PDC
Publication date
2022
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citation
Abstract
En los últimos años, el tráfico entre aplicaciones móviles ha crecido exponencialmente. Así, la tarea de clasificar el tráfico en la red, como puede ser indicar qué aplicación móvil ha generado dicho tráfico, se ha vuelto cada vez más difícil por el elevado uso de protocolos de seguridad para proteger la privacidad de los usuarios y de los datos. Los métodos de clasificación de tráfico basados en el aprendizaje profundo pueden hacer frente a este impedimento y por esta razón el sistema propuesto en este trabajo está basado en autocodificadores, redes neuronales convolucionales y redes neuronales convolucionales gráficas. El conjunto de datos de tráfico con el que se han hecho los experimentos, se ha recopilado durante todo el periodo de realización de este trabajo por un grupo de alumnos y sobre un conjunto de aplicaciones móviles existentes en Android.
In recent years, traffic between mobile applications has grown exponentially. Thus, the task of classifying traffic on the network, such as indicating which mobile application has generated the traffic, has become increasingly difficult due to the high use of security protocols to protect user and data privacy. Deep learning based traffic classification methods can cope with this impediment and for this reason the system proposed in this work is based on autoencoders, convolutional neural networks and graphical convolutional neural networks. The traffic dataset with which the experiments have been done, have been collected throughout the period of this work by a group of students and on a set of existing mobile applications on Android.
In recent years, traffic between mobile applications has grown exponentially. Thus, the task of classifying traffic on the network, such as indicating which mobile application has generated the traffic, has become increasingly difficult due to the high use of security protocols to protect user and data privacy. Deep learning based traffic classification methods can cope with this impediment and for this reason the system proposed in this work is based on autoencoders, convolutional neural networks and graphical convolutional neural networks. The traffic dataset with which the experiments have been done, have been collected throughout the period of this work by a group of students and on a set of existing mobile applications on Android.
Description
Trabajo de Fin de Grado en Ingeniería Informática, Facultad de Informática UCM, Departamento de Ingeniería del Software e Inteligencia Artificial, Curso 2021/2022.