A Trust-Based Malicious Detection Scheme for Underwater Acoustic Sensor Networks

Abstract

Underwater acoustic sensor networks (UASNs) have been widely applied in the fields of maritime and underwater industries and national defense. However, due to the unattended deployment environment of UASNs, the sensor nodes are vulnerable to malicious attacks and are easily compromised to be malicious nodes. In recent years, trust models are proved as an effective and efficient tools for identifying malicious nodes possessing valid identity information. We propose a trust-based malicious identification scheme (TMIS) for UASNs. First of all, the impact of underwater environment on communication trust is quantified, which makes communication trust effectively reflect the behavior of the attacks that cause communication failure such as selective forwarding attacks. Second, communication traffic is exploited to effectively reflect the behavior of the attacks that transmit or receive an abnormal number of packets, such as DOS attack. Third, we train the prediction model with SVM and K-means++ algorithms. Finally, two trust update mechanisms are proposed to cope with the dynamic environment of UASNs and On-Off attacks. The simulation results show that TMIS can effectively identify malicious nodes in complex underwater environment compared to the other three kinds of identification schemes. In particular, the larger the rate of malicious nodes is, the better TMIS performs relatively.