Separation of signals and cross-terms in Wigner distribution using adaptive filtering.
The paper introduces the general approach of separating constant and time-varying spectral terms via the "LMS harmonic canceller", in which the Fourier transform (FT) is applied to the data rather than the data kernel. It is shown that the problem translates into separating two sets (partition over [0,2*]) of spectral components according to their frequency. One set has constant values and includes all frequencies which are integer multiples of 1/N (harmonics), where N is the transform block length. The second set has time-varying components which are non-integer multiples of 1/N (non-harmonic). The employed structure for this task is referred to as the LMS harmonic canceller. The harmonic constant values and the time-varying nonharmonics components correspond, in PWD application to the signals and cross-terms, respectively.
|Main Author:||Amin, Moeness G.|
|Other Authors:||Allen, Fred.|