Satellite Network Security Routing Technology Based on Deep Learning and Trust Management

Zhiguo Liu, Junlin Rong, Yingru Jiang, Luxi Zhang

The conventional trust model employed in satellite network security routing algorithms exhibits limited accuracy in detecting malicious nodes and lacks adaptability when confronted with unknown attacks. To address this challenge, this paper introduces a secure satellite network routing technology founded on deep learning and trust management. The approach embraces the concept of distributed trust management, resulting in all satellite nodes in this paper being equipped with trust management and anomaly detection modules for assessing the security of neighboring nodes. In a more detailed breakdown, this technology commences by preprocessing the communication behavior of satellite network nodes using D–S evidence theory, effectively mitigating interference factors encountered during the training of VAE modules. Following this preprocessing step, the trust vector, which has undergone prior processing, is input into the VAE module. Once the VAE module’s training is completed, the satellite network can assess safety factors by employing the safety module during the collection of trust evidence. Ultimately, these security factors can be integrated with the pheromone component within the ant colony algorithm to guide the ants in discovering pathways. Simulation results substantiate that the proposed satellite network secure routing algorithm effectively counters the impact of malicious nodes on data transmission within the network. When compared to the traditional trust management model of satellite network secure routing algorithms, the algorithm demonstrates enhancements in average end-to-end delay, packet loss rate, and throughput.