Detection of Malicious Nodes in IoT Networks based on Packet Loss using ML

Abstract

Devices can now effortlessly and wirelessly share data with one another through the Internet or other networked systems thanks to newly created technology known as Internet of Things. Despite these advantages, Internet of Things devices are more susceptible to hacker attacks now, which can have unfavorable effects. The Internet of Things ecosystem’s continual proliferation is to blame for this. These invasions could have negative financial and physical effects. A network that automatically configures itself is the Internet of Things. Rogue nodes have the ability to start any number of attacks on this network. For instance, a malicious node may launch a denial-of-service attack by sending a huge quantity of packets at a target node. A threshold-based approach utilizing cutting-edge machine learning techniques is started to locate these malicious nodes in a network. The proposed technique can assist in locating an attacker node by monitoring the path latency and raising an alert if it rises above a predetermined threshold value. The suggested approach will be imitated using the NS2 program. The evaluation and demonstration provided by the proposed technique show that the system in question performs admirably on a number of fronts, including throughput. A test is run across a 500 sq. m area network with 250 packets sent per session and a 1000-byte packet size.