A novel low complexity multichannel Cosine Modulated Filter Bank using IFIR technique for Nearly Perfect Reconstruction

摘要

This paper presents a design of low-complexity multichannel Nearly Perfect Reconstruction (NPR) Cosine Modulated Filter Bank (CMFB). CMFBs are used extensively because of the ease implementation and they have the advantage of high stopband attenuation. But this leads to certain limitations because it would require large number of filter coefficients and longer delays when the number of channels is more e.g. 32-band or 64-band CMFB. Large number of filter coefficients would imply that increased computational complexity. Here the prototype filter is designed using interpolated finite impulse response (IFIR) technique. The stretched factor is chosen in such a way so that computational cost will be reduced and minimizes the value of error parameters. The IFIR approach leads to reduction in stopband energy as well as Side-Lobe-Fall-off-Rate (SLFOR) in magnitude response of the prototype filter. One example has been included to illustrate the effectiveness of the proposed technique over the existing design methods. Here, an earlier approach to prototype filter design based on the Kaiser window is implemented using an IFIR approach. Savings in computational complexity and improvement in performance with respect to the previous works are shown.