Prediction-Assisted Compressed Spectrum Sensing

In compressed spectrum sensing (CSS), true spectrum support is often assumed to be entirely unknown, as is the case in real-world applications. This unknown spectrum support, however, renders the design of the sensing matrix a complex and arduous task. In this paper, we demonstrate that it is possible to obtain certain prior knowledge on the spectrum support, which can then be employed to efficiently design a high-quality sensing matrix. We propose prediction-assisted CSS schemes that incorporate integrated spectrum support prediction methods or rely on external cooperative decisions to update the sensing matrix in accordance with spectrum dynamics. Numerical experiments reveal that the proposed schemes yield significant performance improvements in both single-node and cooperative scenarios.