Decimation IIR filter for oversampled A/D based on multiobjective optimization
Z.P. Ma, Bosco Leung · 2002
Describes a decimation infinite impulse response (IIR) filter design methodology by formulating a multiobjective programming that optimizes both magnitude and phase response. A general condition relating to an approximate linear phase IIR filter design is given. A novel approximation method, called samples-sum-estimation, is introduced to simplify the original objective which provides the flexibility of maximizing the signal-to-noise ratio (SNR) subject to certain anti-aliasing constraints when arbitrary quantization noise power spectral density is presented. Examples show that the proposed design gives SNRs less than 1 dB to the theoretical limit at different decimation ratios, while the group delay displays an approximately constant value.>