Gradient Descent ANN Modelling of RF MEMS Switch for Defense Applications

Radio Frequency Micro Electro Mechanical System (RF MEMS) are progressing as an enabling technology for designing low loss, high performance, power-efficient switches which are essential in designing modern miniaturized phase shifters for defense radar applications. In this paper, an equivalent circuit modelling of RF MEMS shunt switches based on Artificial Neural Network (ANN) method is presented. The RF performance of MEMS based shunt switch are predicted for different Resistance, Inductance, Capacitance (RLC) combinations. The proposed ANN-based circuit modeling to determine better RF performance overcomes the disadvantages of the time-consuming optimization processes involved in EM and circuit simulators. In this work, the proposed Gradient Descent ANN model minimizes the design time of the equivalent circuit model by 88.8% in comparison with circuit simulator that takes nearly 45 minutes. The validation of the proposed ANN model is done by comparing the obtained results with the Advanced Design System (ADS) model and good agreement is achieved.