Minimax design of FIR digital all-pass filters

This paper considers the design of digital all-pass finite impulse response filters in the minimax sense. Our proposed algorithm is a numerical technique to obtain the optimal filter coefficients by using a primal affine scaling variant of Karmarkar's algorithm. Under this approach, we formulate the design problem in terms of minimizing the magnitude and phase errors simultaneously or minimizing the magnitude and group delay errors simultaneously. Computer simulation shows the effectiveness of the proposed design techniques.