Classification of human body activities using low profile wearable antennas.

Monitoring human body activities has long been of interests because of their important applications in human–computer interfacing, virtual control, computer gaming, and medical care industries etc. Traditional methods, such as Doppler radar, depth camera, and wearable sensors, have been extensively studied and they showed good performance at certain scenarios. However, they also have their limitations in terms of their mobility and hardware cost. In this dissertation, a new approach for monitoring body activities is introduced. It is based on the near-field perturbation effect of wearable antenna due to the body movements, which can cause the input impedance of wearable antenna to change. Compared with the above-mentioned radar or physical sensor-based methods, this approach does not only incur low cost and consume low power but also can provide high classification accuracy. Moreover, users do not have to install or set up additional sensors because the antenna from a handheld device can be utilized. In this study, three different kinds of wearable antennas are design, fabricated and tested on the human body. These antennas are the 3D printed folded cylindrical helix, folded cylindrical helix array, and textile patch antenna, not only can theybe used for on-body and off-body communications, but they also can be used to monitor human activities. Several human dynamic experiments were conducted, including finger gestures, head and mouth activities, and arm motions. It is found that different body activities caused the S11 of wearable antennas to vary with unique patterns. Dynamic time warping algorithm classification results showed that more than 90% accuracy can be achieved to classify the 4 different finger motions performed in the experiment.