Radar Tech Comparison for In-Bedroom Activity and Sleep Monitoring

What changed

This paper presents a controlled comparison of three radio frequency (RF) sensing technologies—frequency-modulated continuous wave (FMCW) radar, impulse radio ultra-wideband (IR-UWB), and Wi-Fi sensing—for in-bedroom human activity recognition (HAR) and sleep monitoring. Unlike previous studies that often use differing hardware, datasets, or methodologies, this research employs synchronized recordings from 20 participants across six room layouts, all evaluated using the same convolutional neural network (CNN) architecture. This uniform approach allows for a direct assessment of each technology's performance characteristics.

The study found that IR-UWB achieved the highest cross-subject activity recognition performance, reaching an 89.0% macro F1 score. FMCW radar demonstrated superior generalization capabilities to unseen room layouts, with an 83.8% macro F1 score. For sleep monitoring, all three technologies performed exceptionally well, exceeding 92% macro F1 in unseen environments. The researchers identified a fundamental trade-off between recognition performance and environmental robustness, attributing these differences to variations in range resolution, antenna diversity, Doppler resolution, and spatial information retention among the technologies.

Why it matters for builders

This work offers valuable insights for developers designing contactless, RF-based healthcare monitoring solutions. By providing a direct comparison under identical conditions, the study helps builders understand the strengths and weaknesses of FMCW, IR-UWB, and Wi-Fi sensing for specific use cases. The findings can inform decisions regarding hardware selection, deployment strategies (particularly ceiling-mounted setups), and the development of machine learning models tailored to the unique signal properties of each technology. This is critical for creating effective, privacy-preserving systems for assisted living and remote healthcare.

Practical impact

Developers can leverage these findings to select the most appropriate RF sensing technology for their specific application. For applications requiring high accuracy in recognizing diverse activities across different users, IR-UWB appears to be a strong candidate. If the primary concern is robust performance across various room configurations and environments, FMCW radar shows promise. For sleep monitoring, all tested technologies offer high performance, suggesting flexibility in choice. The study also provides an open synchronized dataset containing FMCW, IR-UWB, and Wi-Fi measurements, enabling reproducible benchmarking and further research into cross-modality RF sensing. Builders can explore this dataset to validate their own models or develop new approaches.

Caveats and source limits

The study focuses on ceiling-mounted radar deployments, which may not fully represent performance in other configurations. While the CNN architecture was kept uniform, the specific hyperparameters and training details are not fully elaborated in the excerpt. The research is presented as a preprint on arXiv and is currently undergoing review for publication in IEEE Access Journal, meaning results may be subject to change. The excerpt does not detail the specific hardware models used for each technology beyond general types (e.g., IWR6843AOP for FMCW, DW3000 for IR-UWB, ZedBoard/openwifi for Wi-Fi), which could be relevant for replication. Pricing and availability of the specific hardware configurations are not discussed.