In this work, design optimization of a varicap diode loaded antenna consisting of four identical rectangular microstrips is presented as a pattern reconfigurable antenna at 5.2 GHz. The microstrips are printed on the front of a FR4 substrate with the dimensions of 40 mm × 25 mm and ε r = 4.6, h = 1.58 mm and probe‐fed via a coupling using a rectangular microstrip line symmetrically placed between them. In first stage, S11 of the antenna are obtained as its real and imaginary parts as continuous functions of geometry of the microstrip components within 3 to 7 GHz using multi‐layer perceptron (MLP) trained and validated by 3D EM simulated data. In order to determine the most suitable (MLP) architecture and training algorithm, 20 different MLP architectures are tested. Then, S11 are optimized with respect to the geometry parameters using differential evolution algorithm and MLP based model. The antenna is prototyped with the optimally selected parameters and measured. From the comparison of simulation and measurement results, it can be observed that the measurement results agree with the simulation results, thus it can be concluded that the proposed antenna is a simple and successful design subject to the design purposes with together its design methodology.

In this article, a simple, accurate, fast, and reliable black-box modeling is proposed for the scattering (S)-parameters and noise (N)-parameters of microwave transistors using the general regression neural network (GRNN) with the substantially reduced measurements and computational cost. In this modeling method, GRNN is employed as a nonlinear extrapolator to generalize the S-data and N-data belonging to only a single bias voltage in the middle region into the entire device operation domain of the bias condition (V DS /V CE , I DS /I C , f) within the shortened human effort. The proposed method is implemented to the modeling of the two transistors BFP640 and ATF-551 M4 as study cases. Thus, comparisons are made with the multilayer perceptrons, trained by the two standard backward propagation algorithms, which are the Levenberg-Marquardt, Bayesian regularization and the 10 data mining methods recently published in the literature using the chosen training data sets in both ınterpolation and extrapolation types of generalization. All the comparisons are achieved using four criteria commonly used in the literature. It can be concluded that GRNN is found to be a fast and accurate modeling method that extrapolates the reduced amount of training data consisting of measured S-parameters and N-parameters at the typical currents of the middle bias voltage to the wide operating range.

The relations between the antennas' geometrical parameters and design specifications usually consist of linear and nonlinear components. Especially with the increase of the requested performance measures, the design procedure becomes much more complex due to the conflicting performance criteria or design limitations. To achieve a design with high performance with feasible design parameters, a fast, accurate, and reliable design optimization process is required. Herein, to have a fast, accurate, and high‐performance capacitive‐feed antenna model to be used in design optimization problems, a modified multi‐layer perceptron (M2LP) model has been proposed. The M2LP is an equivalent convolutional neural network (CNN) model of a standard multilayer perceptron (MLP), where instead of traditional training parameters of MLP, more advanced training parameters of CNN models such as batch‐norm layer, leaky‐rectified linear unit (ReLU) layer, and Adam training algorithm had been used. Furthermore, the M2LP model had been used in a design optimization process and the obtained optimal antenna had been prototyped using 3D printing technology for justification of the proposed M2LP model with experimental results. As can be seen from the results, the proposed M2LP model is a fast, accurate, and reliable regression model for design optimization of microwave antennas.

Reflectarray antennas (RAs) have the ability to combine the advantages of both traditional parabolic reflector and phased array antennas without the need for feed network designs. Microstrip reflectarrays (MRAs) have the advantages of being small size, light weighted, easy to prototyped, high gain, low side-lobe level, and a predetermined radiation pattern. These can be achieved by precise calculation of reflection phase at each RA unit independently with a phase compensation proportional to the distance from the feed. The challenging problem is to have a fast and high accurate unit element to be used in multidimension, multiobjective design optimization. Herein, artificial intelligence algorithms (AIAs) have been used for prediction of reflection phase characterization of an X band MRA unit element with respect to the geometrical design parameters. Firstly, a nonuniform unit RA has been designed in 3D electromagnetic (EM) simulation tool for creating the training validation data sets. Then, the data sets are given to the different types of AIA regression models such as multilayer perceptron, symbolic regression, and convolutional neural network. From the results of the validation data set, it can be concluded that the proposed models have sufficient accuracy that can be used in a computationally efficient design optimization process of a large-scale RA design.

In this work, the simultaneous trade-off relations among the noise figure F, gain G T , input V in , and output V out VSWRs of a microwave transistor operated at a certain (V DS , I DS , f) condition are obtained fast and as accurate as the corresponding analytical results using multiobjective optimization process without any need for expertise on the microwave device, circuit, and noise. Three powerful evolutionary algorithms, cuckoo search, firefly, and differential evolution, are implemented comparatively as a study case to obtain the trade-off relations of a typical low-noise amplifier transistor NE3511S02 for its operation between 9 and 17 GHz at V DS = 2 V and I DS = 10 mA. Finally, differential evolution is found as the most successful algorithm to demonstrate the typical trade-off relations of NE3511S02. It can be concluded that these trade-off relations being obtained by using a signal and noise model of the transistor enable performance database covering all the (F ≥ F min , G T , V in ≥ 1, V out ≥ 1) quadruples with their (Z S , Z L ) termination pairs using solely an evolutionary optimization process. Thus, a small signal transistor can be identified by its performance database to be used in the design optimization of high-performance low-noise amplifiers with the full device capacity.KEYWORDS microwave transistor, multiobjective optimization, competitive evolutionary algorithms, low-noise amplifier (LNA), cuckoo search algorithm, firefly algorithm, differential evolution algorithm | INTRODUCTIONToday, ultra-wideband (UWB) miniature low-noise amplifier (LNA) design with the low-power consumption from the low-level battery is one of the biggest challenges to UWB transceiver integrations. Especially, most of the receivers are hand-held or battery-operated devices; therefore, the very stringent requirements are encountered in the design optimization of an LNA that are mainly a very low-power consumption from a very low-supply voltage, high gain G T , low-noise figure F, and low input V in and output V out standing wave ratios along the UWB.Although the cascode configuration is commonly used in the utilization of the high-performance LNA designs, since it requires bigger than 1 V supply voltage, it is unsuitable to be used in the miniature LNA with the low-voltage applications. For applications requiring very low-supply voltages, a single transistor in the common emitter configuration is typically used.