Multi-person Fitness Assistance via Millimeter Wave

The next generation of WiFi incorporates integrated millimeter-wave technology, which leverages high-frequency radio waves in the millimeter range to enable wireless data transmission. The key advantages of mmWave over traditional WiFi frequencies (such as 2.4 and 5 GHz) are its wide bandwidth and directional characteristics. Recently, there is a growing trend in employing mmWave signals for fitness monitoring, driven by their wide bandwidth and directional characteristics. In this chapter, we present a millimeter-wave-based fitness monitoring system that offers personalized and environment-independent monitoring with reduced training requirements in amulti-person scenario. To address limited training data, we employ a GAN-assisted method that achieves satisfactory performance. Additionally, we develop a domain adaptation training framework to enhance system robustness and enable deployment in new environments with minimal training efforts. Our system utilizes a unique Spatial-Temporal Heatmap feature for personalized workout recognition and incorporates a point-cloud-based method for concurrent multi-person workout monitoring. Extensive experiments demonstrate that our system achieves high accuracy in workout recognition and user identification.