3D human posture reconstruction has rapidly expanded across a wide range of applications. However, current vision-based posture tracking systems face significant challenges, such as privacy concerns and reliance on favorable lighting conditions. To address these issues, recent studies have explored the use of commodity radio frequency signals for 3D human posture tracking, offering a more privacy-preserving and robust solution. Despite these advancements, existing methods struggle to handle scenarios with multiple users in the same environment. In this chapter, we introduce \({m}^3\) Track, which employs a single mmWave radar to simultaneously reconstruct and track the postures of multiple users as they move, walk, or sit. Leveraging sensing signals from a mmWave radar in multi-user environments, \(m^3\) Track first isolates individual users on the radar signals. It then extracts each user’s shape and motion features and reconstructs their 3D postures using a specially designed deep learning model. Afterward, \(m^3\) Track maps the reconstructed 3D postures into 3D space and utilizes a coordinate-corrected tracking method to accurately track user positions. Experiments conducted in real-world multi-user scenarios demonstrate the accuracy and robustness, showcasing its practicality for multi-user 3D posture tracking.