Spectrum Map Construction Method Based on Dynamic Window Size Tensor Ring Low-rank Factors

Abstract

Spectrum maps can model the received signal strength over a geographical region and will play a pivotal role in the intended spectrum management scheme. Traditional spectrum map construction methods cannot fully utilize the spatial-temporal correlation characteristics of observed spectrum data in a time-varying spectrum situation. The computational complexity for real-time scenes is unaffordable, and the current spectrum situation cannot be estimated promptly. To address this problem, we first model the spatial-temporal spectrum data by tensors. Then, based on the low-rank statistical characteristic of the spectrum map, we apply the tensor ring low-rank factors (TRLRF) algorithm to recover the missing spectrum data. Finally, a dynamic window mechanism is introduced to reduce the computational complexity further. The simulation results show that the proposed dynamic window size tensor ring low-rank factors (DW-TRLRF) algorithm yields higher accuracy than other state-of-the-art algorithms with significantly lower complexity.