OPTIMAL MULTILAYER FILTER DESIGN USING REAL CODED GENETIC ALGORITHMS

A novel approach for designing optimal multilayer filters based on a real-coded genetic algorithm is presented. Given the total number of layers in the filter, as well as the electrical properties of the materials constituting each layer, the algorithm iteratively constructs multilayers whose frequency response closely matches a desired frequency response. In contrast to existing iterative techniques, this method does not require a preliminary design using classical techniques. Also, the design procedure is independent of the nature of the multilayer as well as the characteristics of the incident and substrate media. The algorithm is applied to the design of various lowpass and high-pass optical filters, operating between practical terminal conditions. The performance of the resulting designs matches or improves on that for filters that were synthesised using semiclassical techniques.