CNN-based Approach for IoT Intrusion Attack Detection
Atul Kumar, Ishu Sharma
Abstract
Internet of things-based devices is widely used in various applications to ease the usage of technology. The applications of such devices lie in multiple sectors like healthcare, supply chain, warehouses, research and development, smart infrastructure, etc. These devices are equipped with Internet connectivity, sensors, a small battery, and limited bandwidth. The idea behind such devices is to accomplish the goal with the minimum resources. Cyber attackers take advantage of the concept of Internet of Things-based devices by targeting such devices to steal user information from the network. These devices can be attacked through various types of attacks like ransomware, denial of service, malware, data, and identity theft. Nowadays, Artificial Intelligence is widely employed by researchers to detect such attacks, and using optimized modeling for the same task can lead to a secure environment for these devices. In this paper, Convolutional Neural Networks based approach is given to detect internet of Things intrusion attacks. Our proposed approach achieves 0.993 area of Area under the Receiver Operating Characteristic Curve which proves the effectiveness of the proposed technique.