An Optimal Approach in Design of Microstrip-to- SIW Transition Using Bayesian Optimization

Microstrip-to-SIW transition is essential for interconnects in microwave systems. However, conventional design approach of transition optimization highly relies on experience and plenty of simulations, which costs a long time. In this paper, Bayesian-based Optimization method is proposed for the design of microstrip-to-SIW transition, which requires small number of simulations and hence improves design efficiency. The optimization result shows that the number of simulations can be reduced to 39% compared with Pattern Search algorithm, which is the most efficient algorithm aside from Bayesian Optimization mentioned in this paper. The simulation results of the optimized transition shows that an insertion loss is better than 0.7 dB from 16 to 30 GHz.