The IoT (Internet of Things) knows a great expansion with a predicted of 75 billions of IoT devices by 2025 [Cisco]. LPWAN have emerged as one of the most promising IoT wireless technologies. They are appropriate for devices that require small amounts of data over a long-range, at a low cost, while maintaining long battery life. As for any communication systems, security is one of the biggest concerns. In LPWAN, security is currently provided by symmetric-key algorithms such as AES-128 at upper levels. In the case of LoRaWAN, it offers application-level payload encryption and network-level integrity. Nevertheless, this does not shield it against attacks at a lower level such as jammer-type attacks. A jamming attack takes place at the PHY layer. It is an external node that sends unauthenticated packets to every wireless station in the network with the aim of disrupting communications. In this thesis, we focus on LoRaWAN networks under jamming attacks. We address this problem through a holistic approach that encompasses three fundamental axes: •Network modeling: A new mathematical model for LoRaWAN under jamming and an enhancement of a ns3 module for LoRaWAN under jamming. •Performance evaluation of LoRaWAN under jamming: Extensive simulation campaigns. •Jamming countermeasures: (i) a jamming attack detection algorithm using a statistical approach and (ii) a jamming attack detection algorithm using machine learning.