Spectrum Completion Based on HaLRTC

Abstract

With the development of communication technology, the number of communication devices has exploded. In order to utilize spectrum resources rationally, it is very important to analyze the current complex electromagnetic environment. But due to the influence of acquisition equipment, electromagnetic environment noise and other factors, it is often impossible to collect the completed spectrum dataset. This paper mainly studies the method of completing the missing spectrum data, then the completed spectrum data is used to assist the analysis of complex electromagnetic environment. Relying on the correlation of spectrum data in different dimensions, we take advantage of the tensor completion algorithm HaLRTC to design a spectrum tensor completion scheme. Then we compare the differences between the completed data and the original data under different missing ratios, used error judgment indicators including mean absolute error (MAE), mean absolute percentage error (MAPE), root mean square error (RMSE) and equalization coefficient (EC). The four prediction accuracy indicators can complement each other and jointly measure the optimal prediction results. Experiments show that the low-rank tensor completion algorithm based on HaLRTC has better performance in recovering missing spectral data.