CBL - Campus del Baix Llobregat

Projecte llegit

Títol: Seguridad en redes IoT

Estudiants que han llegit aquest projecte:

Director/a: LEÓN ABARCA, OLGA

Departament: ENTEL

Títol: Seguridad en redes IoT

Data inici oferta: 29-01-2025     Data finalització oferta: 29-09-2025


Estudis d'assignació del projecte:
    GR ENG TELEMÀTICA
Tipus: Individual
 
Lloc de realització: EETAC
 
Paraules clau:
Seguretat, IoT
 
Descripció del contingut i pla d'activitats:
Utilitzant dispositius reals i simulats, s'analitzaran diferents protocols IoT des del punt de vista de seguretat, i es proposaran mecanismes per la detecció i prevenció d'amenaces.
 
Overview (resum en anglès):
The Internet of Things (IoT) is experiencing exponential growth, presenting significant security challenges. IoT devices are characterized by their low power consumption and minimal configuration requirements, which is also reflected in the simplicity of the communication protocols they use. This simplicity, which is advantageous from an operational standpoint, comes with serious security limitations. As a result, as the number of connected devices increases, so does the attack surface and the frequency of security incidents.

This paper analyzes the MQTT protocol, widely used in IoT environments due to its lightness and simplicity in message exchange. However, these same characteristics make it a frequent target for attackers, as it lacks built-in protection mechanisms by default. To address this issue, the first objective of the project is to simulate different types of attacks targeting an MQTT broker, as well as benign traffic, in order to analyze the protocol's communication model. All traffic is captured and organized into a dataset, which will be used for the final phase of the work.

Finally, the project integrates simulated traffic generation and the application of machine learning techniques for attack detection. Several models were developed using the collected data and an external dataset. The process consists in data preprocessing, model training, and model evaluation. Subsequently, the model's performance is compared when evaluated with its own test data, collected under the same conditions as the training, against the data collected in the simulated environment for this work. In addition, the study examines how the combination of datasets from different scenarios affects the accuracy of the model.

© CBLTIC Campus del Baix Llobregat - UPC