Attack and Anomaly Prediction in IoT Networks using Machine Learning Approaches

Cyber-attacks are becoming a very common problem we face in IoT networks nowadays. These attacks mainly target the comparatively less powerful IoT and gateway devices so that the attacker can compromise the device much faster. As the IoT infrastructure is getting exponential growth recently, the attacks are also getting the same pace. Based on various studies, almost 90% of attacks targeting IoT networks are Denial of Service (DoS) or Distributed Denial of Service (DDoS). So, like any other aspect, security is an inevitable part of IoT networks and devices because this is now deployed in highly critical areas like space technology, healthcare, smart cities, etc. Understanding these attacks, their behavior pattern, and their defense mechanisms will help to implement a highly safe and secure IoT infrastructure. The system proposed here is a prototype model for detecting the most common DoS attacks at the gateway device itself using the Machine Learning approach. ML approaches can recognize the behavior pattern of the attacks based on the previously trained data and can make efficient predictions with a high accuracy factor. This will help us to take proper countermeasures to protect our IoT network from such potential attacks. The need for a safe and secure IoT ecosystem is an important need for today and especially in the future.