HAL will be down for maintenance from Friday, June 10 at 4pm through Monday, June 13 at 9am. More information
Skip to Main content Skip to Navigation

Modélisation de réseaux IoT hétérogènes à des fins d’évaluation de sécurité

Abstract : The Internet of Things is evolving rapidly, with more and more protocols and objects being deployed, on a large scale or in closed environments. However, with this proliferation of protocols comes new security issues. Indeed, IoT protocols are heterogeneous, specific to particular needs and used in various application domains, making them complex to secure. Several solutions exist to evaluate and improve the security of complex systems, including intrusion testing. In this thesis, we describe a methodology for modeling heterogeneous IoT networks, used as an analysis support to perform penetration testing. We focus this methodology on short-range IoT protocols such as Zigbee, BLE and OS4I. However, its modular approach allows it to be functional for the largest number of IoT protocols, avoiding modification to add a new protocol. For this purpose, we first present a generic approach, based on four criteria, which allows to describe and compare, according to a homogeneous model, several IoT protocols. We then present a classification of attacks on IoT protocols, in three parts, in order to understand and specify the target of the attack and its impact. These abstract and generic models allow us to build a generic structure, which we call generic packet. Our modeling process, composed of four graphs, is based on this generic packet. We propose an iterative scheme to build these graphs, from the graph representing the point-to-point communications of the network, to the one highlighting the applications detected in the network. The generation of each graph requires the use of functions, taking as input several patterns and the previous graph. Finally, we propose an experimentation platform, composed of several objects, mixing proprietary and configurable devices. It allows the evaluation of our modeling methodology under different experimental conditions. In the ideal case, we find that the modeling detects all the applications deployed in the network. The result is also relevant when we degrade our observation through encrypted communications. Indeed, we then find that all applications are detected, despite a higher number of false positives. All the functions and patterns defined in our modeling methodology are implemented in IoTMap, a framework, which we have released as open source.
Document type :
Complete list of metadata

Contributor : Abes Star :  Contact
Submitted on : Tuesday, October 26, 2021 - 2:43:27 PM
Last modification on : Saturday, February 12, 2022 - 3:10:39 AM
Long-term archiving on: : Thursday, January 27, 2022 - 7:33:14 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03404156, version 1


Jonathan Tournier. Modélisation de réseaux IoT hétérogènes à des fins d’évaluation de sécurité. Intelligence artificielle [cs.AI]. Université de Lyon, 2021. Français. ⟨NNT : 2021LYSEI018⟩. ⟨tel-03404156⟩



Record views


Files downloads