Publication: Análisis de trafico en dispositivos móviles mediante técnicas de aprendizaje profundo no supervisado
Loading...
Official URL
Full text at PDC
Publication Date
2022-06
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
En la actualidad, debido al auge de las tecnologías móviles es posible realizar tareas cotidianas como trabajar, tomar clases en lineal, conversar con familiares o amigos y jugar de manera remota desde un smatphone. Sin embargo, mezclar estos ámbitos de nuestra vida resulta atractivo a un ciberdelincuente debido a que en muchas ocasiones en un mismo equipo móvil se tienen aplicaciones de uso exclusivo de una empresa como correo electrónico o repositorios digitales y estas a su vez interactúan con aplicaciones de ocio o potencialmente peligrosas para la seguridad de cualquier organización. Estas acciones pueden propiciar desde una exfiltración de datos o documentos confidenciales, hasta que el equipo sea comprometido por un ciberdelincuente y tenga vía libre a los datos internos de la empresa. Existen diferentes formas de minimizar el riesgo, una de las comunes en cualquier organización es el monitoreo de la red, sin embargo, el uso de algoritmos de cifrado de los datos de las aplicaciones esta no es suficiente ya que la mayoría de las aplicaciones móviles utilizan robustos algoritmos de cifrado para ocultar sus datos lo cual dificulta la labor de los analistas de Red. En la actualidad existen soluciones que permiten visualizar el contenido de las comunicaciones cifradas mediante módulos para debilitar los algoritmos de cifrado a versiones vulnerables, lo cual podría repercutir en la integridad de los mensajes de red. Por lo anterior en el presente trabajo se muestra una herramienta capaz de clasificar trafico de aplicaciones móviles mediante su comportamiento en la red utilizando técnicas no supervisadas de Deep Learning.
Today, due to the growth of mobile technologies is possible to do daily activities such as work from home, take online classes, talk with family or friends and playing remotely from a smartphone. Nevertheless, mixing this things of our lives, result attractive to a cybercriminal due to in many occasions, in one same mobile device there is the possibility to have applications of exclusive use for an enterprise such as email or digital repositories and at the same time, these interact with leisure applications potentially dangerous for the security of any organization. These actions can produce data filtration or confidential documents, until the device is compromised by a cybercriminal and has a way to the internal data of an enterprise. There exist different ways of minimizing risks, one of the most common ones in any organization is network monitoring , nevertheless, the use of data encryption algorithms of the applications is not enough due to the majority of the applications using robust encryption algorithms to hide the data so is difficult for the network analysts work. Nowadays, there exist solutions that allow to show the encrypted communication content by modules to weaken the encryption algorithms to vulnerable versions, that can take part in the integrity of network messages. Taking this into account this work presents a tool for being able of classify mobile applications traffic by the network behaviour using non supervised techniques.
Today, due to the growth of mobile technologies is possible to do daily activities such as work from home, take online classes, talk with family or friends and playing remotely from a smartphone. Nevertheless, mixing this things of our lives, result attractive to a cybercriminal due to in many occasions, in one same mobile device there is the possibility to have applications of exclusive use for an enterprise such as email or digital repositories and at the same time, these interact with leisure applications potentially dangerous for the security of any organization. These actions can produce data filtration or confidential documents, until the device is compromised by a cybercriminal and has a way to the internal data of an enterprise. There exist different ways of minimizing risks, one of the most common ones in any organization is network monitoring , nevertheless, the use of data encryption algorithms of the applications is not enough due to the majority of the applications using robust encryption algorithms to hide the data so is difficult for the network analysts work. Nowadays, there exist solutions that allow to show the encrypted communication content by modules to weaken the encryption algorithms to vulnerable versions, that can take part in the integrity of network messages. Taking this into account this work presents a tool for being able of classify mobile applications traffic by the network behaviour using non supervised techniques.
Description
Trabajo de Fin de Grado en Ingeniería Informática, Departamento de Ingeniería de Software e Inteligencia Artificial, Curso 2021/2022.