Detection of selective-edge packet attack based on edge reputation in IoT networks

Abstract

The Internet of Things (IoT) is widely used in environmental monitoring, smart home and other fields, and has become a research hotspot in recent years. Due to its distributed nature, IoT is vulnerable to various attacks. In traditional packet attack, malicious nodes will indiscriminately attack packets. The existing detection algorithms can detect the attack by observing the overall behaviour of nodes. In this paper, we introduce an advanced attack named selective-edge packet attack, in which malicious nodes only attack packets sent to specific neighbours. Due to selectively attack packets, the attack is more covert than the traditional packet attack, which makes it difficult for the existing detection algorithms to detect. To detect selective-edge packet attack, we propose machine learning-based detection framework (MDMK) and machine learning-based detection algorithm (MDA) based on the framework, which uses regression algorithm and clustering algorithm to evaluate the reputation of communication links and nodes, and detects malicious nodes accordingly. To further improve the detection performance, machine learning-based detection algorithm with enhancement (MDAE) is designed by optimizing the routing path. The experimental results demonstrate that compared with the existing detection algorithms, the accuracy of MDA and MDAE in different situations is improved by around 7%–30% on average.