Font Size: a A A

Design of IIR filters with canonical signed-digit (CSD) coefficients using genetic algorithms

Posted on:2004-12-03Degree:M.A.ScType:Thesis
University:University of Windsor (Canada)Candidate:Liang, LiFull Text:PDF
GTID:2468390011468725Subject:Engineering
Abstract/Summary:
In this thesis, an optimization method is used in the design of 1-D IIR digital filters, doubly complementary filter pairs and 2-D IIR digital filters. This method uses genetic algorithm (GA) to minimize the mean square error between the desired and the designed filter responses in order to calculate the coefficients of the filter's transfer function.; The 1-D IIR filters are designed using cascade structure, and constraints are imposed on the coefficients of the denominator polynomials to ensure stability. In the one-dimensional case, the method is also used to design doubly complementary filter pairs, in which case two filters that are both all-pass complementary and power complementary are designed at the cost of one filter. Based on the same stability criterion, an optimization method is presented for the design of 2-D IIR digital filters with non-separable numerator and separable denominator transfer functions.; The advantage of the proposed method is that it produces filters with Canonical Signed-Digit (CSD) coefficients, which not only eliminates the quantization process in digital filter design but also make the designed filter more efficient for high speed DSP applications. Design examples of 1-D IIR filters, doubly complementary filter pairs and 2-D IIR filters are provided in order to demonstrate the usefulness of the presented method.
Keywords/Search Tags:IIR filters, 1-D IIR, Method, Coefficients
Related items