| Autor: |
Chaitanya, Pranav, Sethuraman, S., Kanthamani, S., Roomi, S. Mohamed Mansoor |
| Předmět: |
|
| Zdroj: |
Microsystem Technologies; Jan2024, Vol. 30 Issue 1, p105-116, 12p |
| Abstrakt: |
Radio Frequency (RF) Nano Electro Mechanical Switches (NEMS) using Carbon Nanotube (CNTs) are preferred for electronic switching due to its high stiffness, switching speeds and ultra-low mass over several GHz range. CNT switch for microwave and millimeter wave applications utilizing a maximum of three CNTs with fixed starting positions of CNTs from zero level offset to obtain better isolation (S21) of the switch in DOWN State has already been studied. The motive of this paper is to obtain better RF performance in terms of return loss (S11) and isolation loss by modelling the switch in DOWN-state by varying the random positioning of CNTs, interelement spacing between CNTs, number of CNTs and frequency is predicted and reported using XG—Boost approach. Time-consuming optimization methods to predict the S-parameters in customized Electromagnetic (EM) simulators such as HFSS (High Frequency Structure Simulator) have been resolved using the proposed XG-Boost regression approach. In this paper, the proposed XG-Boost approach minimizes the switch design time with random placement of CNTs by 99.81% compared to the conventional EM simulator which requires about 120 h for 400 simulations in an Intel-Core processor (2.50 GHz) with 8 GB RAM. To validate the results, they are compared to the HFSS model, and a high degree of agreement is found. Mean Square Error (MSE) of the switch in the DOWN- state after comparing the EM simulation values and XG-Boost regression values is 6.832e−03. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Complementary Index |
| Externí odkaz: |
|
Full text from SpringerLink