Spectrum Situation Awareness for Space-Air-Ground Integrated Networks Based on Tensor Computing.

Area of Science:

• Network Engineering
• Data Science
• Signal Processing

Background:

• Space-air-ground integrated networks (SAGINs) generate vast multidimensional heterogeneous big data.
• Traditional methods struggle with the complexity and scale of SAGINs data.
• Effective spectrum situation awareness is critical for SAGINs performance.

Purpose of the Study:

• To investigate the application of tensor computing for spectrum situation awareness in SAGINs.
• To develop a novel spectrum awareness scheme leveraging tensor data models and computations.
• To explore the utility of tensor eigenvalues in characterizing spectrum data and enabling advanced sensing.

Main Methods:

• A multidimensional tensor data model was designed for SAGINs.
• Tensor decomposition and completion were employed for data dimensionality reduction and missing data imputation.
• Tensor eigenvalues were calculated and their distributions analyzed.
• Spectrum sensing schemes were designed based on tensor eigenvalue distributions and hypothesis testing.

Main Results:

• Tensor computing effectively handles multidimensional heterogeneous big data from SAGINs.
• Tensor eigenvalues provide insights into intrinsic data correlations, crucial for situation awareness.
• A novel tensor-computing-based spectrum awareness scheme was proposed and validated.
• Simulation results demonstrated the feasibility of spectrum awareness using tensor eigenvalues.

Conclusions:

• Tensor computing offers a powerful paradigm for spectrum situation awareness in SAGINs.
• Tensor eigenvalues serve as a robust statistical indicator for spectrum awareness.
• This work opens new avenues for applying tensor theory in practical communication networks.