Coupling matrix synthesis by optimization with hybrid cost function based on Hausdorff distance and wavelet D4 transform

This paper proposes a new hybrid cost function used in filter coupling matrix synthesis. Coupling matrix synthesis is a very important step in microwave filter design process and can be done in a way of optimization which needs a well designed cost function. The first term of proposed cost function is defined by the Hausdorff distance between template and optimized sets of zeros and poles extracted from a filter scattering characteristics and coupling matrix formulae. The second term is the Euclidean distance between compressed Daubechies D4 wavelet transform of designed and optimized filter scattering parameters. The experiments showed a very high performance of the proposed hybrid cost function definition in comparison to the methods used nowadays in coupling matrix optimization technique.