A review of optimization structures and algorithms in the design of microwave sensors

The purpose of this paper is to provide an analytical review of optimization structures and algorithms in the design of microwave sensors, with a focus on novel artificial intelligence approaches. The main research problem is that the performance of microwave sensors—especially planar resonant sensors—heavily depends on the resonator geometry, field concentration region, substrate properties, sample loading method, and parameter tuning algorithm. Therefore, manual design and trial-and-error methods are insufficient for simultaneously achieving high sensitivity, adequate quality factor, small dimensions, and fabrication robustness. The research method is review-analytical, based on an examination of recent scientific literature on SRR, CSRR, DGS, microfluidic, metamaterial, surrogate model, and machine learning algorithm-based sensors. This review shows that complementary resonators and metamaterial structures offer relative advantages in terms of electric field concentration in the sensing region and are more suitable for permittivity measurements. The use of algorithms such as genetic algorithms, particle swarm optimization, or Bayesian optimization alongside these structures reduces design time and improves the accuracy of material property estimation. However, it should be noted that a significant gap still exists between simulation results and the performance of fabricated prototypes, and fully automated design, despite recent advances, has not yet reached sufficient maturity.

Keywords:

Microwave sensor , Optimization , Resonator , Machine learning , Inverse design