A new Delayless Subband Adaptive Filter based on Discrete Cosine Harmonic Wavelet Transform (CHWT)

摘要

This paper proposes a new delayless subband adaptive filter (SAF) based on Discrete Cosine Harmonic Wavelet Transform (CHWT). The subband signals derived using CHWT are of real nature due to DCT. This enables to obtain the estimate of the overall plant impulse response in the DCT domain and hence is simple compared to that of difficult stacking used in delayless SAF realized by DFT filter banks. Further the aliasing due to filtering is absent as subband signals are derived in DCT domain and hence critical sampling is sufficient. This does not suffer from eigen value spread, due to over sampling used to avoid aliasing. The wavelet nature in addition to handling nonstationarity provides the perception. Hence, the new algorithm is simple and provides good convergence rate for signals with large eigenvalue spread and of nonstationary nature.