On Antenna Q-factor Characterization with Generative Adversarial Networks
- Resource Type
- Authors
- Jayakrishnan Vijayamohanan; Oameed Noakoasteen; Arjun Gupta; Manel Martínez-Ramón; Christos G. Christodoulou
- Source
- 2020 IEEE International Symposium on Antennas and Propagation and North American Radio Science Meeting.
- Subject
- Distribution (number theory)
Computer science
Q factor
Perspective (graphical)
Parameter distribution
Antenna (radio)
Characterization (mathematics)
Solver
Algorithm
Generative grammar
Computer Science::Information Theory
- Language
This paper introduces a novel way to reproduce antennas with Q-factor within a pre-determined threshold using Generative Adversarial Networks (GAN), a class of artificial intelligence algorithm. Instead of optimizing the Q-factor using a conventional optimization techniques, a GAN is trained to learn the distribution on desirable parameter vectors of the antenna and its Q-factor as obtained from a full wave solver. The trained GAN is subsequently used to generate antenna parameters from the learned distribution with a predicted Q-factor. The predicted antenna parameters are imported to CST and simulated using the parameters generated by GAN and the Q-factors are analyzed. The results obtained provides an unique perspective to learning the correlation between antenna parameter distribution and its Q-factor. Simulation results are provided to explain this approach of reproducing antennas with a low Q-factor.