An Edge-Computing-Enabled Trust Mechanism for Underwater Acoustic Sensor Networks
Jiaxin Du, Guangjie Han, Chuan Lin
Abstract
The trust model has become a promising mechanism to detect anomalous sensor nodes and guarantee security in underwater acoustic sensor networks (UASNs). A trust model is realized by collecting different trust metrics and transforming them into trust values as reliability measurements. However, trust collection and calculation exhibit high latency in the traditional UASN structure, and anomalous nodes cannot be identified quickly. In addition, despite the existing extensive research on trust management, the adaptive adjustment of the weights for different trust metrics has not been solved yet. To tackle these challenges, we integrate edge computing technology into UASNs, and regard the autonomous underwater vehicle (AUV) as the edge device to provide low-latency trust modeling services. We propose an attention-based trust model for edge-computing-enabled UASNs, which we name ATrust. The attention mechanism can adaptively adjust the contributions of different metrics to trust modeling, since the attention mechanism highlights the valuable trust metrics related to the final trust evaluation results. In addition, we present the deployment strategy for AUVs to reduce the probability of being attacked by analyzing the characteristics of the thermocline in the ocean.