Recovering Missing Values from Corrupted Historical Spectrum Observations for Dependable Spectrum Prediction

Spectrum prediction is a promising technology to infer spectrum state from historical spectrum observations, by exploiting the inherent correlations and regularities among them. Due to the common existence of missing values and anomalies in the real-world spectrum observations, spectrum prediction with incomplete and corrupted historical observations has caused extensive concern. In this paper, we aim to tackle the challenging problem on how to accurately and efficiently recover the missing values from corrupted historical spectrum observations with which dependable spectrum prediction can be performed. To this end, we first formulate a hankelized time-structured spectrum tensor that can naturally preserve both spectral and temporal dependencies among the historical spectrum observations. Then we model the spectrum data recovery as a tensor completion problem by exploiting its latent low-rank structure and sparse anomaly property. To efficiently solve the optimization problem, we design a robust online spectrum data recovery algorithm based on the alternating direction method. Numerical results demonstrate that the proposed algorithm outperforms state-of-the-art schemes and confirm its effectiveness for dependable spectrum prediction.