A novel discrete coefficient IIR digital filter design based on a lattice structure
Hong Bo Xu, Chaogeng Huang, Feng Ye, Tao Hong · 2014
Based on a “tapped numerator” and “injected numerator” hybrid lattice structure, a novel discrete coefficient IIR digital filter design is proposed in this paper. This lattice structure possesses much improved finite word length properties such as very low parameter sensitivity and very uniform signal powers across signal nodes, so it is very suitable for FWL implementation. There are three sets of parameters in this lattice structure, and in [8], only one set of parameters was optimized in a discrete coefficient space. In [9], all sets of parameters were optimized together over a discrete power-of-two coefficient space but the approach fails when the filter order becomes high due to the fact that the searching space gets very big and the nonlinearity of the cost function increases greatly. Our proposed method can efficiently obtain high order discrete coefficient IIR filter with lower complexity, better satisfactory frequency response and very uniform signal power ratio.