Abstract
Tracking interpersonal distances is essential for real-time social distancing management and ex-post contact tracing to prevent spreads of contagious diseases. Bluetooth neighbor discovery has been employed for such purposes in combating COVID-19, but does not provide satisfactory spatiotemporal resolutions. This paper presents ImmTrack, a system that uses a millimeter wave radar and exploits the inertial measurement data from user-carried smartphones or wearables to track interpersonal distances. By matching the movement traces reconstructed from the radar and inertial data, the pseudo identities of the inertial data can be transferred to the radar sensing results in the global coordinate system. The re-identified, radar-sensed movement trajectories are then used to track interpersonal distances. In a broader sense, ImmTrack is the first system that fuses data from millimeter wave radar and inertial measurement units for simultaneous user tracking and re-identification. Evaluation with up to 27 people in various indoor/outdoor environments shows ImmTrack's decimeters-seconds spatiotemporal accuracy in contact tracing, which is similar to that of the privacy-intrusive camera surveillance and significantly outperforms the Bluetooth neighbor discovery approach.
Introduction: The Contact Tracing Challenge
Effective contact tracing for infectious diseases requires "decimeters-seconds" resolution—the ability to know exactly how close individuals were and for how long. Current solutions generally fall into two categories, both of which have significant drawbacks. Bluetooth Neighbor Discovery (BND) is widely used but suffers from high latency and inaccurate distance estimation due to signal multipath and attenuation, often failing to detect brief contacts. Conversely, device-free sensing like camera surveillance is highly accurate but raises severe privacy concerns.
Millimeter-wave (mmWave) radar offers a promising middle ground, providing high-resolution 3D data without identifying faces. However, it faces an "anonymity problem": while it can track moving objects, it cannot inherently link a specific track to a specific person's identity. ImmTrack bridges this gap by fusing the high-precision tracking of radar with the pseudo-identities provided by a user's smartphone IMU data.
Methodology: Bridging the Modality Gap
ImmTrack employs a sophisticated multi-stage fusion process to achieve simultaneous tracking and re-identification.
IMU-Assisted mmWave Tracking
To prevent the radar from merging two people into a single cluster during close proximity, ImmTrack utilizes Recursive Kalman Filters (RKF). These filters predict user positions in subsequent frames, initializing a motion-aware $k$-means clustering algorithm. This ensures that individual clusters remain distinct even during near-contacts. Further, a custom deep neural network, mmClusterNet, extracts shape and motion features from radar point clouds to maintain consistent tracking across frames.
Cross-Modality Trajectory Association
The core innovation lies in matching movement "trace maps." The system converts 2D trajectories from both the radar and the smartphone's IMU (derived via dead reckoning) into image-based trace maps where pixel intensity represents dwell time. A Siamese Neural Network then compares these maps to associate a radar track with a specific smartphone ID. If trace maps are too similar (e.g., people walking side-by-side), the system employs a Gait Analysis Fallback, comparing radar Doppler spectrograms against IMU acceleration patterns.
Evaluation and Key Results
The system was evaluated in diverse environments, including a sports hall, an outdoor playground, and a four-day deployment in a 300 $m^2$ office space with up to 27 participants.
- High Precision: ImmTrack achieved an average interpersonal distance tracking error of just 22 cm.
- Robust Contact Detection: For identifying contacts within a 1-meter threshold, the system reached 90% precision and 94% recall.
- Low Latency: Compared to BND, ImmTrack reduced discovery latency by up to 80 seconds, enabling the detection of fleeting encounters.
- Computational Efficiency: ImmTrack provides accuracy comparable to camera-based systems while requiring significantly less computational power, making it ideal for edge deployment.
Conclusion
ImmTrack demonstrates that fusing mmWave radar with IMU data can provide the high-resolution tracking necessary for effective contact tracing without compromising individual privacy. By turning movement patterns into unique identifiers, the system overcomes the inherent anonymity of radar sensing, offering a scalable and efficient solution for managing public health in shared spaces.
Summary updated