Hybridization of Evolutionary Algorithm with Yule Walker Method to Design Minimal Phase Digital Filters with Arbitrary Amplitude Characteristics

In this paper a method of design minimal phase digital filters with arbitrary amplitude characteristics is presented. The filters designed using proposed method can be directly implemented in the hardware. Filter coefficients obtained using proposed method are immune on rounding error. Because these coefficients are created in required bit word length (based on numerical format which is used in given DSP system). The proposed method is based on connection of evolutionary algorithm and Yule Walker method. Due to this hybridization, digital filters are designed faster (than using only evolutionary algorithm) and possess much better properties (then using only Yule Walker method). Digital filters obtained using proposed method are ready to be implemented in DSP system without any additional errors. The four minimal phase digital filters (with coefficients in Q.15 format) and with arbitrary amplitude characteristics are designed using proposed method. The results obtained using proposed method are compared with results obtained using other techniques taken from literature.