Internet traffic detection and classification has been thoroughly studied in the last decade, but this is still a hot topic as regards the Internet of Things (IoT), a communication paradigm that is going to involve different aspects of our daily life. As a consequence, researchers started applying traditional methods for traffic classification also to the traffic flows coming and addressed to smart devices. In this paper, we created a large integrated dataset of IoT traffic flows, coming from four different network scenarios, in order to have a benchmark for future research. Moreover, we used this dataset to test the effectiveness of a deep learning network model, made of different hidden layers, and we compare its outcomes with the ones obtained through traditional machine learning approaches, demonstrating the superiority of our deep learning architecture in both a binary and multinomial classification.

IoT Attack Detection with Deep Learning Analysis

Riccardo Pecori
;
Luca Veltri
2020-01-01

Abstract

Internet traffic detection and classification has been thoroughly studied in the last decade, but this is still a hot topic as regards the Internet of Things (IoT), a communication paradigm that is going to involve different aspects of our daily life. As a consequence, researchers started applying traditional methods for traffic classification also to the traffic flows coming and addressed to smart devices. In this paper, we created a large integrated dataset of IoT traffic flows, coming from four different network scenarios, in order to have a benchmark for future research. Moreover, we used this dataset to test the effectiveness of a deep learning network model, made of different hidden layers, and we compare its outcomes with the ones obtained through traditional machine learning approaches, demonstrating the superiority of our deep learning architecture in both a binary and multinomial classification.
2020
978-1-7281-6927-9
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11389/30149
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 32
  • ???jsp.display-item.citation.isi??? ND
social impact