Spectrum Representation Based on STFT
Xuebao Wang, Ying Tao, Wei Tian
Abstract
A new spectrum representation method based on short time Fourier transform (STFT) is proposed. The formula of new spectrum representation is derived base on energy cumulant of short time Fourier transform (EC-STFT), which indicates that EC-STFT has the spectrum characteristics. Simulations on the linear frequency modulation (LFM) signal show that the EC-STFT spectrum is closer to the ideal spectrum curve than FT spectrum. During calculating EC-STFT, a time frequency domain iterative mean threshold (TFD-IMT) denoising method is presented to remove the addictive white Gaussian noise (AWGN), by which the EC-STFT spectrum has better anti-noise capacity than FT spectrum. Theoretical analyses and simulations verify the advantages of the EC-STFT over FT in conditions of low SNR.