Interference-Term Elimination for Wigner Distribution Using a Frequency Domain Adaptive Filter

Abstract

In a nonstationary signal analysis, it has been clarified that the discrete Wigner distribution, WD, is the most suitable distribution. The use of the WD is, however, often complicated by the occurrence of interference terms. Thus, we proposed a modified WD, called the RID (reduced interference distribution), and verified that the RID could markedly suppress the occurrence of interference terms in both the auto-and the cross-WD. In this study, we developed another method which makes use of a block adaptive filter to eliminate interference terms and evaluate them quantitatively. In this method, the least mean square, LMS, algorithm is applied in the frequency domain. Simulated signals of a chirp type and an acoustic signal of a concert hall are analyzed using spectrograms, WD, RID and the new method. From comparisons among the results, it is concluded that the best result can be obtained using the new method proposed in this study.