Optimized Human Motion Estimation Through Extended Kalman Filter in MmWave Radar

This letter presents a novel approach for human motion estimation using millimeter-wave (mmWave) radar, integrating the Extended Kalman Filter (EKF), Convolutional Neural Network (CNN), and Bidirectional Long Short-Term Memory (BiLSTM) networks. The proposed method addresses key challenges in mmWave radar systems, such as frame failures and background noise, by leveraging EKF to enhance tracking robustness and suppress noise. The CNN-BiLSTM classifier captures spatial and temporal features to accurately classify four human motions. Experimental results demonstrate the system's effectiveness, achieving high accuracy rates compared to existing literature. This work advances mmWave radar-based motion detection by introducing a novel 3D state-space model and a hybrid EKF-CNN-BiLSTM framework, offering significant improvements over traditional signal processing and deep learning techniques.