A Fast Identification Method of Shortwave Radio Stations Based on Sparse Component Analysis

Abstract

Aiming at the problem of identification of a large number of radio stations in the high frequency (HF) band, a fast identification method based on sparse component analysis, in which high-speed spectrum scanning data are used to separate and identify multiple stations on each channel, is proposed. Taking into account the adverse effects of the shortwave time-varying channel fading on the radio signals, utilizing the periodicity of the radio signals, a sparse component analysis algorithm based on time feature clustering (TFC-SCA) is proposed. The algorithm combines the time features with the amplitude features for clustering and realizes the accurate estimation of the mixing matrix under fading channel conditions. In addition, based on the clustering results, the algorithm projects the signals to the vectors from the origin to the clustering centers to remove the noise introduced by the time-varying channel fading. In simulation experiments with different duty cycles and different periods, the correlation coefficients of TFC-SCA are closer to 1 than clustering based sparse component analysis (C-SCA) and fast independent component analysis (FastICA), providing a good solution to the problem of separation and identification of shortwave radio stations.