본문 바로가기
  • Home

A Weighted LSTM-Based Prediction of Electrical Characteristics in Feedback Field-Effect Transistors

  • Journal of Internet of Things and Convergence
  • Abbr : JKIOTS
  • 2025, 11(3), pp.67~71
  • Publisher : The Korea Internet of Things Society
  • Research Area : Engineering > Computer Science > Internet Information Processing
  • Received : February 26, 2025
  • Accepted : May 21, 2025
  • Published : June 30, 2025

Seung-won Son 1 Woo Sola 2

1국립부경대학교 전자정보통신공학부
2국립부경대학교

Accredited

ABSTRACT

This study develops a prediction model using a small long short-term memory(LSTM) network to analyze the electrical characteristics of feedback field-effect transistors (FBFETs) based on Technology Computer-Aided Design (TCAD) simulation data. As FBFETs exhibit hysteresis characteristics due to abrupt current changes at latch-up and latch-down transitions, it is difficult to achieve high prediction accuracy by using traditional methods such as multi-layer perceptron (MLP). To address this problem, we applied a parameter-efficient LSTM model, achieving R²= 0.998 and RMSE = 2×10⁻⁸, by optimizing data preprocessing and incorporating a weighting strategy based on latch-up and latch-down currents and voltage values that reflect the electrical characteristics of FBFETs. The trained model parameters were extracted to develop a compact model, enabling Simulation Program with Integrated Circuit Emphasis (SPICE) simulations. This prediction model facilitates the design and optimization of FBFET-based memory and computing devices, contributing to improved performance and reduced power consumption in IoT applications.

Citation status

* References for papers published after 2023 are currently being built.