A novel approach of design and analysis of fractal antenna using a neurocomputational method for reconfigurable RF MEMS antenna

A mathematical neural approach/artificial neural network (ANN) for the design of a swastika-shaped reconfigurable antenna as a feedforward side is proposed. Further design parameter calculations using the reverse procedure of the above method is presented. Neural network computational is one of the optimization methods that could be considered to improve the performance of the device. In this paper, the proposed planar antenna up to the 2nd iteration is simulated using finite element method-based HFSS software. The developed ANN algorithm method allows the optimization of the antenna to be carried out by exchanging repetitive simulations and also provides reduced processing times while still retaining great accuracy as compared to traditional mathematical formulation. The simulated S-parameter (return loss) results of the proposed antenna are verified with the ANN and show good agreement. Furthermore, for proof of concept, the above proposed antenna as well as a swastika-shaped reconfigurable antenna (2nd iteration) with radio frequency microelectromechanical system switches are fabricated and tested using a vector network analyzer. The results presented here show that the antenna works well in the frequency range of 1.5 to 6.5 GHz and resonates on multiple bands. The novelty of the approach described here offers ease of designing the process using the ANN algorithm, maintaining the miniaturization of antenna size, multiband behavior, and utility of the antenna in the mobile terminal.