Robust Time-Aware Streaming Tensor Completion Algorithm for Space-based Spectrum Situation Map Construction

With the help of low Earth orbit (LEO) satellites, space-based spectrum monitoring systems can obtain electromagnetic spectrum situation map (SSM) over a large geographical scope. However, since the collection of measurements is limited by the predetermined satellites orbits, SSM needs to be constructed through incomplete measurements instead of being directly observed through LEO satellites. Besides, due to the influence of cosmic radiation and harsh ground-to-satellite transmission environments, spectral measurements inevitably contain unexpected anomalies. Moreover, the existing batch-based SSM construction methods do not consider the dynamic changes in spectrum situation over time, and the construction of SSM at each time slot requires recalculating from the scratch, making it impossible to implement the dynamic evolution of SSM. Regarding the above issues, we propose a robust time-aware online streaming tensor completion algorithm, formulating the SSM construction as a high-order incomplete streaming tensor completion problem and reconstructing complete SSM by utilizing the time-frequency-space correlation and temporal properties in real-world spectral measurements. Numerical results show that the proposed algorithm can detect anomalies with high accuracy and exhibits better construction performance than state-of-the-art algorithms under different missing and anomaly injection rates.