Body Sensor Design for Unattended, Untethered Deployment
Brantley, Jeffrey, Computer Engineering - School of Engineering and Applied Science, University of Virginia
Lach, John, Department of Electrical and Computer Engineering, University of Virginia
The field of body sensor networks (BSNs) has emerged with the potential for improving patient outcomes and quality of life. Wirelessly-tethered BSN devices for motion assessment and other sensing modalities have been successfully deployed in BSN research studies conducted on patient populations. However, as a potential replacement--at least in part--for often-unreliable patient self-report and frequent, but short, in-clinic visits, BSNs must be able to operate unattended and untethered from a wireless connection in order to provide longitudinal data collection in the naturalistic setting of a user's daily routine.
However, this kind of operation presents practical challenges for BSN firmware. Being untethered from a PC, the firmware must take on increased responsibility otherwise handled in software. Reliable and robust operation become more critical as there is no attending operator to notice and quickly correct any problems that arise. Finally, the firmware must manage its own persistent storage, maintaining a correct, consistent state in spite of unexpected resets, unexpected removal and replacement of the flash storage medium, and eventual degradation of the medium itself.
Additionally, an untethered BSN device must intelligently manage its scarce energy resources despite unexpected daily loads. A tradeoff arises between achievable battery life and the fidelity level of the data collected, and this tradeoff can vary day-to-day with user behavior and the corresponding load on the device. One approach for predictively managing this tradeoff based on personal activity profiles is presented, with a focus on duty cycle adaption for a motion-capture BSN device. A simulation study is performed based on actual daily walking activity profiles obtained from three human subjects wearing Fitbit(R) trackers over several months. Simulation results show improvements with this method over statically setting the duty cycle for constant power consumption with respect to ideally setting the duty cycle based upon a priori knowledge of activities of interest throughout the system lifetime.
MS (Master of Science)
activity profiling, battery lifetime, body sensor network
All rights reserved (no additional license for public reuse)