Application of evolutionary algorithm to design minimal phase digital filters with non-standard amplitude characteristics and finite bit word length

Abstract

In this paper an application of evolutionary algorithm to design minimal phase digital filters with non-standard amplitude characteristics and with finite bit word length is presented. Four digital filters with infinite impulse response were designed using the proposed method. These digital filters possess: linearly falling characteristics, linearly growing characteristics, nonlinearly falling characteristics, and nonlinearly growing characteristics, and they are designed using bit words with an assumed length. This bit word length is connected with a processing register size. This register size depends on hardware possibilities where digital filter is to be implemented. In this paper, a modification of the mutation operator is introduced too. Due to this modification, better results were obtained in relation to the results obtained using the evolutionary algorithm with other mutation operators. The digital filters designed using the proposed method can be directly implemented in the hardware (DSP system) without any additional modifications.