| Abstract |
loT Devices and Sensors are increasingly being used in various sectors, including industrial architecture and smart cities. The security of sensor networks has gotten much attention recently. Introducing new loT networks has shifted the paradigm to vast Internet of Things (loT) networks. loT networks provide new applications and opportunities for WSN networks and open up new possibilities. However, the security of these networks must be taken carefully, and a new approach to loT network security must be proposed. The new options give way to new types of attacks, and the security of these networks becomes very critical. Routing in loT networks uses a different protocol for communication for low-power and lossy networks (RPL). These loT routing protocols are vulnerable to various attacks, such as rank attacks, hello flood attacks, and wormhole attacks. There are many techniques to prevent these attacks, but implementing security techniques in these networks may degrade the network's performance. Traditional security services such as secrecy, authentication, integrity, authorization, non-reputation, and availability make it harder to provide security in low-energy devices. This paper proposes a new machine-learning model for attack detection in loT networks based on parameters like packet received ratio, number of messages between nodes, and packet length or duration. Data is collected by a simulation network and analyzed using the proposed model, which gives good accuracy. © 2024 IEEE. |
