Machine Learning Technique to Improve an Impedance Matching Characteristic of a Bent Monopole Antenna

Round 1

Reviewer 1 Report

The manuscript concerns the optimization of an ideal monopole antenna using a machine learning technique.
I find the manuscript well-written and the results are sound, which are of interest to the scientific readership. The methods and results are presented in a simple and easy-to-understand way.

 

I only have a few comments:

• How does your work differs from:
  Y. Sharma, H. H. Zhang and H. Xin, "Machine Learning Techniques for Optimizing Design of Double T-Shaped Monopole Antenna," in IEEE Transactions on Antennas and Propagation, vol. 68, no. 7, pp. 5658-5663, July 2020, doi: 10.1109/TAP.2020.2966051.
• You consider an ideal monopole antenna (infinite ground and thin wire). I think your results would be much more interesting, if you have a more realistic antenna with finite ground plane, wire thickness, metallic losses etc. Please comment on your design from a practical point of view. E.g. can you make a perfect 90° bend? How will this bend be affected by vibrations and movements?
• The wire bends will cause discontinuities in the fields, where a finite conductivity will have greater influence on the antenna performance. Are these effects taken into account?
• There are no experimental results, which is a bit curious, since the antenna is simple and standard microwave equipment can be used for the measurements. The work would be much more solid, if experimental results were included.
• In the introduction, it is written that the radiation pattern is omni-directional, whereas in fact it is not. Also, the monopole antenna is not used that much nowadays due to its limited directivity and vertical size.
• What does NEC stands for?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors introduce a machine learning (ML) technique to optimize the bending point in space and the wire length from the point of view in the impedance matching for the design of the monopole antenna constructed by bending wire. The authors insist that this study can be an alternative to the design time by EM simulation and the difficulty of the design process due to the need for modeling.

However, in order to replace the EM simulation method, which is already verified and generalized the analysis and design method, the validity, reliability, ease, and scalability of the proposed method should be presented with sufficient measurement and verification results.

I cannot agree on the scientific soundness, significance, effectiveness, and practical meaning of the proposed method. The proposed method has no meaning in the antenna design and cannot replace the EM simulation method in the field.

The ML technique must learn with the verified data. These data must be obtained from EM simulation or measurement data as mentioned in lines 81-82 of the manuscript. Wire modeling is very simple in the ANSYS HFSS 3D EM simulator, and it consumes fewer resources and time than those of the learning process in the ML technique. Including the parameter extraction and optimum design process, the proposed method cannot be easy as compared to the conventional design using EM simulation. The authors should clearly present the reason why the proposed study is necessary and the rationale for it numerically.

The optimum design proposed in the manuscript for impedance matching is practically useless because there are important parameters in the antenna design, such as bandwidth, antenna gain, and beam pattern, and these should generally be considered simultaneously during optimization during the design process.

The authors should not compare the proposed method with other ML techniques but with other methodologies in the antenna design including the optimization techniques.

The authors should provide evidence that the proposed method is superior to the parameter optimization provided by EM simulation.

The equivalent circuit model presented in Figure 6 is very vague. It is difficult to find the basis for modeling a vertical wire as a series resonator and a horizontal wire as a shunt resonator. Moreover, it is not clear how the equivalent circuit is added in the model only by the difference in parameters.

In order to claim that the proposed method is useful in antenna design, it is absolutely necessary to compare the results using the proposed antenna design method with the results using conventional design methodology, especially by using the measurement results.

In the realization of the antenna, it is very difficult to implement the bending angle of the wire antenna with a resolution of 0.1 degree. And the design parameters presented by the authors cannot be practically implemented.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The bent wire monopole atnenna by using machine learning technique to obtain a good impedance matching at 1 GHz is presented in this paper. The results show that the error is less than 1%. This paper should be accepted for publication.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf