Research Area

I am conducting research on ISAC (Integrated Sensing and Communication) systems across both “sensing-oriented waveforms” (e.g., FMCW, PMCW) and “communication-oriented waveforms” (e.g., OFDM, OTFS). In my view, the integration of communication and radar should go beyond simple spectrum sharing and pursue the efficient joint use of spatial, temporal, and frequency resources. This is particularly important in vehicular technologies with high-mobility and multipath-rich scenarios, where the limitations of conventional sensing become more pronounced. To address these challenges, I focus on waveform-dependent ISAC system design considering RF propagation characteristics. I am currently focusing on extending these techniques to more realistic system models and channel environments. To support this, I am also developing and validating prototype systems that directly transmit and receive these waveforms using software-defined radios (SDRs) and perform signal processing based on real measurement data.

To overcome the complexity of real-world environments and the nonlinear limitations inherent in RF signal processing pipelines, I am actively pursuing learning-based signal processing research for multi-modal sensors. In particular, many downstream radar tasks and perception software tasks often lack fully interpretable analytical models and rely heavily on heuristic design. My research focuses on replacing these empirical approaches with effective AI-driven methods. In addition, raw-data-level multimodal sensor processing is a key technology for Software-Defined Vehicles (SDVs), which requires a deep understanding of RF domain. Building on RF propagation and statistical characteristics, My researches have been performed using real measurements from radar, LiDAR, GNSS, IMU, cameras, ultrasonic sensors, and Wi-Fi, and I am currently extending this work by integrating it with deep-learning-based multimodal perception architectures. My ongoing research includes reliability-aware dynamic weighting, label transfer, and joint representation learning, with the goal of developing robust perception systems that remain effective under diverse sensing conditions and real-world uncertainties.