In this communication, a novel artificial neural network (ANN) model is proposed to describe the antenna performance with various parameters. In this model, three parallel and independent branches are involved for three different performance parameters. Meanwhile, a data-classification technique of support vector machine is also included to classify geometrical variables into the proper categories. Once the geometrical variables are input, the ANN model can simultaneously obtain S-parameter, gain, and radiation pattern from the independent branches. The validity and efficiency of this proposed model are confirmed with a Fabry-Perot resonator antenna example.
5 April 2021 in IEEE Transactions on Antennas and Propagation
To improve the convenience and efficiency of antenna design, in this article, a novel inverse artificial neural network (ANN) model is proposed in which antenna performance indexes are set as the input and corresponding geometrical variables are set as the output. To solve the multiobjective problem of antenna modeling, the first part of the ANN model involves three parallel and independent branches for S -parameter, gain, and radiation pattern, and the second part outputs a final predicted result. Once the training is completed, the proposed inverse model can provide antenna geometries directly without being repetitively called by an optimization algorithm. Compared with the inverse extreme learning machine and five state-of-art forward models, the proposed model uses a small number of training samples and directly obtains the satisfying values of geometrical variables without any optimization process.
4 October 2019 in IEEE Transactions on Antennas and Propagation
This communication presents a novel compact broadband circularly polarized (CP) implantable patch antenna for biomedical applications in 2.4 GHz Industrial, Scientific, and Medical (ISM) band. By employing slots on the patch plane and shorting method, the whole size (including the superstrate) of the designed antenna can be enormously reduced to 9.8×9.8×1.27 mm3, which is equivalent to 0.0784λ0 × 0.0784λ0 × 0.0102λ0 (λ0 is the free-space wavelength at 2.4 GHz). Under the premise of keeping a slotless ground, not only a good size-reduction is realized, but also the impedance bandwidth and the axial ratio (AR) bandwidth are greatly improved. A simulated -10 dB impedance bandwidth of 21.5% with a 3 dB AR bandwidth of 15.8% is acquired in a skin phantom. The antenna also exhibits good robustness to different thicknesses of biocompatible coating. The influence of different body phantoms is discussed to evaluate the sensitivity of the proposed antenna. The experiments are carried out in both skin-mimicking gel and pork, and the respective measured impedance bandwidths are 25.9% and 25.7%. The CP purity is experimentally validated by comparing the communication link levels in two orthogonal polarization directions. The maximum specific absorption rate (SAR) value satisfies the IEEE standard safety guidelines.
An efficient knowledge-based artificial neural network (KBANN) is proposed, and it is used for the design of circularly polarized (CP) lens antenna in this article. In this KBANN, forward neural network (FNN) and inverse neural network (INN) are included. In this model, INN is the major component to predict the antenna structure parameters. As multiple performance indices are required, INN requires a large number of training samples to deduce complex mapping relationship. To solve this problem, FNN is introduced to provide prior knowledge for INN. FNN generates a huge training dataset for INN training, and then the trained INN can directly output the geometric parameters by feeding the target electromagnetic responses as input. This article solves the problem of multiple performance indices in antenna design, and a CP lens antenna with wideband, good axial ratio, and high gain is designed and fabricated to verify the effectiveness of the KBANN model.
3 July 2017 in IEEE Transactions on Antennas and Propagation
A method based on the microstrip patch mode and reconfiguration technique is analyzed, studied, and applied to wide-angle scanning phased arrays. First, a summative evaluation of microstrip patch modes available for wide-angle scanning or near endfire has been made. Second, to validate this new method, a multilayer microstrip patch element is established and its resonant modes TM01 and TM20 are excited. Third, a reconfigurable technique is used to reconfigure the above resonant modes and a wide joint 3-dB beam coverage is obtained. Fourth, a prototype of 1 × 4 uniform linear phased array with the above reconfigurable elements is proposed. The prototype operates at 5.8 GHz and its main-beam direction can scan from -75° to +75° in the elevation plane with a gain fluctuation less than ±1 dB. Finally, in order to reduce the sidelobe levels, a hybrid particle swarm optimization algorithm is used to optimize the excitation amplitudes and phases. The measured results show an excellent wide-angle scanning performance, which validates the effectiveness of the proposed method.