1. Improved downlink transmission throughput by 20% by designing a light-weight CNN-based physic-inspired CSI upsampling approach which was filed as a patent2. Improved throughput by 18% by developing UL-CSI-assisted PMI upsampling algorithm and considering standardized Type II precoding feedback in 5G-NR

1. Achieved 2-degree angle resolution by designing a low-complexity subspace angle estimation for automotive mmWave MIMO radars.2. Saved 95% of computation time by proposing a low-complexity MUSIC algorithm3. Improved SINR by 6 dB by designing Wi-Fi 802.11ad inteference suppression algorithm with propotype built with USRP4. Improved 50% of estimation error by designing a ST-LSTM-based trajectory prediction learning model for vehicle motion prediction and tracking as compared to… Show more 1. Achieved 2-degree angle resolution by designing a low-complexity subspace angle estimation for automotive mmWave MIMO radars.2. Saved 95% of computation time by proposing a low-complexity MUSIC algorithm3. Improved SINR by 6 dB by designing Wi-Fi 802.11ad inteference suppression algorithm with propotype built with USRP4. Improved 50% of estimation error by designing a ST-LSTM-based trajectory prediction learning model for vehicle motion prediction and tracking as compared to ConvLSTM prediction method Show less

Reduced 10% of the feedback cost for channel estimation state feedback by excluding redundant states validated through the C++ implementation of 4G LTE on the Link-Level Simulator.