Computationally Efficient Design Optimization of Compact Microwave and Antenna Structures

Abstract

Miniaturization is one of the important concerns of contemporary wireless communication systems, especially regarding their passive microwave components, such as filters, couplers, power dividers, etc., as well as antennas. It is also very challenging, because adequate performance evaluation of such components requires full-wave electromagnetic (EM) simulation, which is computationally expensive. Although high-fidelity EM analysis is not a problem for design verification, it becomes a serious bottleneck when it comes to automated design optimization. Conventional optimization algorithms (both gradient-based and derivative-free ones such as genetic algorithms) normally require large number of simulations of the structure under design, which may be prohibitive. Considerable design speedup can be achieved by means of surrogate-based optimization (SBO) where a direct handling of the expensive high-fidelity model is replaced by iterative construction and re-optimization of its faster representation, a surrogate model. In this chapter, we review some of the recent advances and applications of SBO techniques for the design of compact microwave and antenna structures. Most of these methods are tailored for a design problem at hand, and attempt to utilize its particular aspects such as a possibility of decomposing the structure. Each of the methods exploits an underlying low-fidelity model, which might be an equivalent circuit, coarse-discretization EM simulation data, and approximation model, or a combination of the above. The common feature of the presented techniques is that a final design can be obtained at the cost of a few evaluations of the high-fidelity EM-simulated model of the optimized structure.