Design and Analysis of Microstrip Filters Using Artificial Neural Network

Abstract

The present communication systems are going wireless at fast pace and need to have better quality of components, which are being used in wireless communication systems. Filters are one of the primary components used for selection of wanted signals to have communication with better performance. Microwave filters are commonly used for passing the desired band of frequencies and rejecting other frequencies for various modern microwave applications such as satellite-communication, Radar and mobile communication etc. The work of this thesis was focused on filter design, mathematical modeling of the filter designs along with performance parameters like scattering parameters, VSWR and group delay. In this thesis work, designing of microstrip filters were done using IE3D Electromagnetic simulator, which is based on method of moments and after obtaining sufficient amount of IE3D E.M. simulation of filters (low pass, high pass, band pass and band stop filters) data for training the neural network model was prepared to improve accuracy of the design. After training of the design using data obtained from simulation results; The Scattering parameters and accurate dimensions of that filter can be predicted using neural network model obtained. It eases our computational effort, which was the main motive behind this work. The microstrip line geometry of the microstrip filter is obtained as a function of input variables, which are height of the dielectric material (h), dielectric constants of the substrate material (and#949;r), and the mid-band or cut-off frequency (fc), using artificial neural network techniques. Similarly, in the analysis of filter, the Scattering parameters of the filter is obtained as a function of the microstrip line dimensions (W, L), height of the dielectric substrate (h), and dielectric constants of the material (and#949;r). Thus, the forward and reverse sides of the problem will be defined as synthesis and analysis for the microstrip filter respectively.