ST-GNN-Powered Compression for Latency Reduction in Hybrid Satellite-Terrestrial Networks

Hybrid Satellite-Terrestrial Communication Networks (HSTCNs) have emerged as a critical infrastructure for modern communication, integrating the capabilities of satellite and terrestrial networks to meet the increasing demands for connectivity and reliability. However, enhancing data transmission efficiency in such heterogeneous networks remains challenging, particularly concerning latency. This research aims to address this challenge by proposing a lossless compression method leveraging custom Spatio-Temporal Graph Neural Network (ST-GNN) architectures to reduce data transmission latency in HSTCNs. The primary objectives of this study are twofold: first, to evaluate the compression rates achieved by different ST-GNN-based network traffic compression methods, and second, to conduct a comparative analysis of data transmission latency over HSTCNs based on the compression of traffic networks. By exploiting the inherent spatio-temporal characteristics of communication data, ST- GNNs offer a promising approach to compress network traffic while minimizing latency effectively. The empirical assessment demonstrates the effectiveness of the proposed ST-GNN-based compression methods in achieving competitive compression rates compared to existing techniques. Furthermore, the comparative analysis highlights the potential of ST-GNNs in improving data transmission efficiency and reducing latency in HSTCNs.