Modeling Antenna Radiation Using Artificial Intelligence Techniques: The Case of a Circular Loop Antenna

Modeling Antenna Radiation Using Artificial Intelligence Techniques: The Case of a Circular Loop Antenna

Theodoros N. Kapetanakis (Department of Electronic Engineering, Hellenic Mediterranean University, Greece), Ioannis O. Vardiambasis (Department of Electronic Engineering, Hellenic Mediterranean University, Greece), Melina P. Ioannidou (Department of Information and Electronic Engineering, International Hellenic University, Greece), and Antonios I. Konstantaras (Department of Electronic Engineering, Hellenic Mediterranean University, Greece)
DOI: 10.4018/978-1-7998-4879-0.ch007
OnDemand PDF Download:
No Current Special Offers


The forward and the inverse problem of a thin, circular, loop antenna that radiates in free space is modeled and solved by using soft computing techniques such as artificial neural networks and adaptive neuro fuzzy inference systems. On the one hand, the loop radius and the observation angle serve as inputs to the forward model, whereas the radiation intensity is the output. On the other hand, the electric field intensity and the loop radius are the input and output, respectively, to the inverse model. Extensive numerical tests indicate that the results predicted by the proposed models are in excellent agreement with theoretical data obtained from the existing analytical solutions of the forward problem. Thus, the employment of artificial intelligence techniques for tackling electromagnetic problems turns out to be promising, especially regarding the inverse problems that lack solution with other methods.
Chapter Preview


Artificial intelligence incorporates a vast family of biologically inspired techniques that attempt to imitate human cognitive skills. The artificial neural networks (ANNs) and the fuzzy logic systems, such as the adaptive neuro fuzzy inference system (ANFIS), belong to the aforementioned family of techniques. The former simulate the complex networks of the neurons that are found in the human central nervous system (Haykin, 1999), whereas the latter combine the learning abilities of neural networks with fuzzy logic (Jang, 1993).

During the last few decades, a great variety of electromagnetic (EM) problems have been treated by applying methods based on artificial intelligence (Zhang & Gupta, 2000), (Christodoulou & Georgiopoulos, 2001), (Mishra, 2001). The number of pertinent works is huge. Indicative applications include the modeling and design of microwave components and circuits (Devabhaktuni et al., 2001), (Zhang et al., 2003), as well as the design and optimization of antennas (Choudhury et al., 2015). The prediction of the performance of various types of antennas by using ANNS and/or ANFIS has drawn strong attention, recently. For example, the resonant frequencies of L-shaped compact microstrip antennas have been determined by applying ANNs and ANFISs (Kayabasi et al., 2014), the return-loss characteristics and the radiation patterns of pyramidal and corrugated horn antennas have been obtained by implementing ANFIS-based models (Pujara et al., 2014) and the return-loss performance of a planar inverted-F antenna has been predicted by using an ANFIS (Gehani & Pujara, 2015). Furthermore, the radiation characteristics of a short dipole array, by applying radial basis function (RBF) ANNs (Mishra et al., 2015), and those of a circular loop antenna by implementing multilayer perceptron (MLP) ANNS and ANFISs (Kapetanakis et al., 2012a), (Kapetanakis et al., 2018a), have been calculated.

A more challenging task may be the implementation of artificial intelligence techniques in order to solve inverse EM problems. The latter focus on estimation of the properties of the scatterer or the radiator from information contained in the EM field and obtained either from measurements or from analytical/numerical calculations. ANNs and the Finite Element Method (FEM) have been combined in (Low & Chao, 1992) to treat inverse EM problems, whereas an approach based on Hopfield ANNs has been presented in (Elshafiey et al., 1995). Inverse ANΝ modeling has been applied for the purpose of microwave filter design (Kabir et al., 2008) and for the inversion of a transmitarray database, i.e., the estimation of the element parameters from the transmission coefficients of the antenna array (Gosal et al., 2016). Moreover, several ANN configurations (Kapetanakis et al., 2018b) and an ANFIS (Kapetanakis et al., 2012b) have been implemented in order to model and solve the inverse EM problem of a thin, circular, loop antenna that radiates in free space.

The radiating circular loop antenna is used herein as an example in order to investigate the potential of certain artificial intelligence techniques to treat EM problems. Although the first efforts to solve the equations that describe the behavior of loop antennas occurred many decades ago, the study of such antennas draws strong interest even nowadays. The motivation may be attributed to the fact that there exist several problems that have not been solved completely yet, for example those related to the radiation of thick loops. Moreover, the emerging meta-material theory was a boost for the evolution and application of nano-scaled rings (McKinley, 2019).

The basis for the study of loop antennas was established more than one century ago by Pocklington, who obtained a solution, in the form of a Fourier series, for the current on a closed loop excited by a plane wave (Pocklington, 1897). Several years later, Hallen considered a driven loop but his Fourier series solution for the current contained a singularity; thus, the results of his method were limited to small loops (Hallen, 1938). Since then, a lot of great scientists have contributed to the formulation and solution of the analytical equations that govern the radiation of a loop antenna. In the 1950s and 1960s several researchers, by complementing one another, managed to find a well behaved solution for the current and the impedance of a driven thin, circular loop in the form of a convergent, infinite Fourier series (Storer, 1956), (Wu, 1962), (King, 1969). Presentation of a detailed analytical history of loops is beyond the scope of this chapter; a comprehensive essay on this subject may be found in (McKinley, 2019).

Key Terms in this Chapter

Metamaterial: Manmade material with properties that cannot be found in materials encountered in nature, for example material with negative refractive index.

MSE: Mean square error.

RMSE: Root mean square error.

ANFIS: Adaptive neuro fuzzy inference system.

MAE: Mean absolute error.

GR: Generalized regression.

Artificial Intelligence: Machines or techniques that attempt to imitate human cognitive skills.

Nanotechnology: Techniques, methods and processes that manipulate materials with dimensions of the order of nanometers.

AE: Absolute error.

Radiation Intensity: The power radiated from an antenna per unit solid angle, in the far-field.

MRE: Mean relative error.

Gaussian Function: A function that has a characteristic, symmetric graph which resembles the shape of a bell.

MLP: Multilayer perceptron.

Complete Chapter List

Search this Book: