Detection Protocol for the Five Variants of Selective Forwarding Attack in the Internet of Things

Abstract

RPL (Routing Protocol for Low power and lossy networks), the most widely used routing protocol in the Internet of Things, has numerous security vulnerabilities. These make it particularly susceptible to selective forwarding attacks. This work aims to add a scalable, fast, and accurate detection process to RPL for the five variants of this attack, regardless of the type of data collected. To this end, each node is constantly evaluated by its neighbors. Packets routing is modeled as a maximum flow problem, which allows for the prediction of maximum throughput and average delivery delay. By comparing these indicators to the node’s actual performance, each neighbor estimates a trust level. The node's final status is determined through feedback from the neighbors, following an approach inspired by Dempster-Shafer theory. Simulation showed thattheproposed scheme is precise, energy-efficient, and outperforms similar recent state-of-the-art contributions.