DDoS attack detection in IoT devices using ANN based IDS in a simulated environment

The Internet has expanded a lot in terms of its geographical coverage as well as its accessibility, since its inception in the late 20th century. With its rise, the number of devices connected with it are also increasing, especially with the advent of Internet of Things. IoT devices are being utilized in a variety of applications and nowadays, they form the base of crucial Infrastructure of a country as well, due to which they are becoming prime targets to being hacked and attacked for malicious purposes. DoS is one such attack which disrupts the accessibility to the server, restricting access to it for the legitimate users. When it is carried out simultaneously by multiple devices (bots) then the attack is called the DDoS (Distributed denial of service) attack. Due to the increasing vulnerability of IoT devices to such attacks and the impact of them in today's digitized world, it has gathered attention of researchers in this problem domain in the recent past. In this paper, we have focused on the role of Deep Learning for this task. We have proposed an Invasion Detection System based on Machine Learning which leverages the ANN (Artificial Neural Networks) for combating such attacks. The ANN is trained on IP traces data to classify safe and hostile packets, and based on this knowledge it is able to detect a DDoS attack. This framework is tried out in a simulated IoT Environment and is able to achieve significant accuracy.