1. Cyberspace-Electronic Warfare Simulations

This study focuses on maximizing jamming performance through the utilization of state-of-the-art machine learning techniques instead of traditional jamming methods. The transmitter predicts channel conditions through spectrum sensing, while the jammer employs adversarial machine learning to mimic the transmitter's model. Due to the inherent difficulty in obtaining data in the field, data augmentation is frequently employed as a means of addressing this challenge. In this study, GANs will be utilized for performing data augmentation.

2. Wireless Occupancy Monitoring

This project focuses on energy-saving solutions within buildings, particularly addressing the challenge of monitoring occupancy for energy conservation. Traditional methods, like PIR sensors and cameras, are costly and unsuitable for complex indoor environments. In this paper, we introduce WiSOM, a self-adaptive occupancy detection system that uses WiFi's channel state information (CSI). We thoroughly evaluated WiSOM's performance under various indoor conditions, including multipath effects, different intensities and instances of activities, and wall absorption. The results show WiSOM's high detection rate, resilience to CSI variations, and significant improvement over recent baselines in real-house scenarios, making it a promising solution for efficient energy management in buildings.

Related Project

국방과학연구소 - 사이버 전자전 송신신호 모의분석 모델링 기법 연구 (2022.06.01 ~ 2024.05.31)