IEEE MTT-S 2008 International Microwave Symposium Technical Program

Session: TH4B	Advanced Techniques for CAD
Chair:	Arvind Sharma, Northrop Grumman
Co-Chair:	Andreas Cangellaris, University of Illinois at Urbana-Champaign
Abstract:	Advanced design optimization techniques using rigorous analytical and numerical procedures are presented. Space mapping, EM sensitivity, and neural network techniques provide robust practical solutions for microwave design.

TH4B-01	Rigorous Computer-Aided Design of Coaxial/Circular Antennas with Semi–Spherical Dielectric Layers
1523	C. Tomassoni¹, M. Mongiardo¹, P. Russer², R. Sorrentino¹, ¹Universtity of Perugia, Perugia, Italy, ²Technische Universität München, Munich, Germany
	We present a new rigorous formulation for studying antennas made by a central circular (or elliptical) waveguide, and a coaxial line; the circular waveguide being the inner part of the coaxial line. The antenna may be covered with spherically stratified dielectric layers for improving radiative properties. The use of spherical transmission lines provides the rigorous network representation of the antenna. It is shown that, by expressing the radiated field in terms of spherical modes, is advantageous for numerical computations and yields rigorous network representations. The proposed formulation moreover follows the same procedure used for discontinuities in closed waveguides, thus unifying the treatment of electromagnetic field propagation for guided and radiated wave problems.

TH4B-02	Efficient Electromagnetic Optimization Using Self-adjoint Jacobian Computation Based on a Central-node FDFD Method
1585	X. Zhu, A. Hasib, N. K. Nikolova, M. H. Bakr, McMaster University, Hamilton, Canada
	We propose a sensitivity solver for frequency-domain analysis engines based on volume methods such as the finite-element method. Our sensitivity solver computes S-parameter Jacobians directly from the field solution available from the electromagnetic simulation. The computational overhead is a fraction of that of the simulation itself. It is independent from the simulator’s grid, system equations and discretization method. It uses its own finite-difference grid and a sensitivity formula based on the frequency-domain finite-difference (FDFD) equation for the electric field. It computes the S-parameter gradients in the design parameter space through a self-adjoint formulation which eliminates adjoint system analyses and greatly simplifies implementation. We use our sensitivity solver in gradient-based optimization of filters. We achieve drastic reduction of the time required by the overall optimization process. All examples use a commercial finite-element simulator.

TH4B-03	A General EM-Based Design Procedure for Single-Layer Substrate Integrated Waveguide Interconnects with Microstrip Transitions
1767	J. E. Rayas-Sanchez¹, V. Gutierrez-Ayala², ¹ITESO, Tlaquepaque, Mexico, ²Intel - Guadalajara Design Center, Tlaquepaque, Mexico
	We propose in this work a general procedure to efficient EM-based design of single-layer SIW interconnects, including their transitions to microstrip lines. Our starting point is developed by exploiting available empirical knowledge for SIW. We propose an efficient SIW surrogate model for direct EM design optimization in two stages: first optimizing the SIW width to achieve the specified low cutoff frequency, followed by the transition optimization to reduce reflections and extend the dominant mode bandwidth. Our procedure is illustrated by designing a SIW interconnect on a standard FR4-based substrate.

TH4B-04	Adaptive Space Mapping with Convergence Enhancement for Optimization of Microwave Structures and Devices
1004	S. Koziel², J. W. Bandler¹, Q. S. Cheng¹, ¹McMaster University, Hamilton, Canada, ²Reykjavik University, Reykjavik, Iceland
	A novel space mapping algorithm is presented that adaptively adjusts the type of space mapping surrogate model used in a given iteration, based on the approximation and generalization capabilities of the model, its ability to satisfy the design specifications, as well as convergence properties of the iterative optimization process. The new technique allows us to avoid a wrong choice of space mapping surrogate which might lead to poor performance of the space mapping algorithm. No extra fine model evaluations are necessary as the assessment process uses only data emerging naturally during the optimization procedure. The performance of the method is verified using microwave design optimization examples and is compared with the previously published adaptive space mapping algorithm.

TH4B-05	Tuning Space Mapping: A Novel Technique for Engineering Design Optimization
1017	J. Meng¹, S. Koziel², J. W. Bandler¹, M. H. Bakr¹, Q. S. Cheng¹, ¹McMaster University, Hamilton, Canada, ²Reykjavik University, Reykjavik, Iceland
	We introduce a tuning space mapping (TSM) technology for microwave design optimization. For the first time, we formulate the novel TSM concept and show how it relates to the standard space mapping methodology. The new method is based on a so-called tuning model that is created using engineering expertise and knowledge of the design problem, but also utilizes the efficiency of space mapping for translating the adjustment of the tuning parameters into relevant updates of the design variables. We illustrate our approach through optimization of a high-temperature superconducting (HTS) filter.

TH4B-06	Robust Training of Microwave Neural Network Models Using Combined Global/Local Optimization Techniques
1763	H. Ninomiya¹, S. Wan², H. Kabir², X. Zhang², Q. Zhang², ¹Shonan Institute of Technology, Fujisawa, Japan, ²Carleton University, Ottawa, Canada
	This paper presents a new technique for training microwave neural network models. The proposed technique combines quasi-Newton algorithm with a recent global optimization algorithm called Particle Swarm Optimization (PSO). The quasi-Newton process for searching optimal solutions is incorporated into PSO to speed up convergence during local search, while the PSO performs global search avoid being trapped in local minima of training. The overall algorithm iterates between quasi-Newton and PSO. Neural network training for waveguide and microstrip examples are presented to demonstrate the proposed algorithm. The proposed algorithm shows more accurate and robust training results than the conventional gradient based technique and the conventional PSO.

If you encounter any problems, please contact