|
Abstract
|
IoT device classification has a variety of applications. These applications include creating a more efficient infrastructure, improving security, and better data analysis. IoT device classification helps operators gain a complete view of the devices connected to the network and manage them based on type, performance, and security priorities [5]. In addition, device classification plays an important role in preventing and combating security threats. IoT devices are attractive targets for cyberattacks due to their connected and sometimes vulnerable nature. Accurate identification and classification of these devices allows operators to apply appropriate security policies to each type of device and provide a quick response in case of security breaches. At first glance, it may seem that IoT devices can be identified through MAC addresses and DHCP protocols, but this method faces serious challenges. For example, many IoT device manufacturers use network cards with invalid OUI prefixes that do not provide information about the device or that MAC addresses can be spoofed, and many devices do not provide their requested hostnames in the DHCP protocol. Users can also change the hostname, which makes accurate identification more difficult. For these reasons, relying on the DHCP infrastructure is not a reliable solution for accurate identification of IoT devices on a large scale. Therefore, classifying IoT devices is a challenging issue. One of the methods for classifying IoT devices is to use the features of network traffic flows [6]. In this research, we try to provide a practical solution for classifying IoT devices by extracting these features and classifying them through neural network-based approaches.
|