Enhancing Satellite System Reliability through Bistable RF MEMS Switch: An Optimization Approach using XGBoost model

Radio Frequency Micro Electro Mechanical System (RF MEMS) switches offer unparalleled reliability and swift response times with compact design, making them an ideal choice for ensuring faultless redundancy in critical satellite systems. This paper proposes a Crab leg Bistable RF MEMS switch to perform redundancy functions in satellite applications. To optimize the proposed switch, eXtreme Gradient Boosting (XGBoost) model is presented, which is proven to be an efficient Machine Learning (ML) model. The proposed XGBoost model was developed based on both direct and indirect approaches to determine the RF and physical characteristics in both stable states of the switch. The XGBoost model overcomes the time-consuming EM optimization problem. In this work, the proposed model aims to significantly reduce the duration required for designing the switch. In comparison to the Electro-Magnetic (EM) simulator, which demands nearly 50 minutes for each iteration, the proposed model reduces the design time by 90%. The effectiveness of the developed ML model is validated by comparing predicted results with those obtained from the HFSS model, resulting in the achievement of favorable convergence.