Research Topics

• Application of Quantum-inspired algorithm to massive MIMO detection
  • Implemented the optimization algorithm inspired by quantum computing for large MIMO detection. Developed GNU Radio block based on this algorithm and evaluated using software-defined radios (SDRs).
• Reinforcement Learning-based Adversarial Attack on Neural Networks for Wireless Human Activity Recognition
  • Developed an attack scheme to degrade deep learning-based human activity recognition (HAR) using Wi-Fi signals by manipulating pilot signals from user devices. To overcome the limitation of the gradient-based attack method, which requires knowledge of future channel information, proposed a generative adversarial imitation learning (GAIL)-based algorithm that imitates the gradient-based attack without these assumptions, successfully degrading HAR model performance.
• Modulation Classification for Intelligent spectrum sensing
  • Participated in the IARPA-funded intelligent spectrum sensing project, leading a modulation classification part for Wi-Fi 6 and 5G signals. The project aims to identify anomaly signals Developed a preprocessing algorithm for deep learning-based classification without preamble or control channel information. Implemented PHY layer channels for 5G and Wi-Fi 6 using MATLAB toolboxes and demonstrated that our proposed algorithm reliably identifies modulation of practical OFDM signals using SDR.