Center for Embedded Network Sensing

Recent advancement in microsensor technology permits miniaturization of conventional physiological sensors. Combined with low-power, energy-aware embedded systems and low power wireless interfaces, theses sensors now enable patient monitoring in home and workplace environments in addition to the clinic. Low energy operation is critical for meeting long operating lifetime requirement; an energy-aware wearable system is therefore particularly beneficial to adaptively profile and manage energy utilization. Furthermore, important challenges appear as some of these important physiological sensors, such as electrocardiographs (ECG), introduce large energy demand (because of the need for high sampling rate and resolution) and limitations (due to reduced convenience of user wearability). Energy usage of the distributed sensor systems may be reduced by activating and deactivating sensors according to real-time measurement demand as well as energy consumption characteristics. Our results show that with proper adaptive measurement scheduling, an ECG signal from a subject may be needed for analysis only at certain times, such as during or after an exercise activity. This demonstrates that autonomous systems may rely on low energy cost sensors combined with real time computation to determine patient context with high certainty diagnostics and apply this information to properly schedule use of high cost sensors (e.g. ECG sensor systems).

We have implemented a wearable system based on standard widely-used handheld computing hardware components. This system relies on a new software architecture and an embedded inference engine developed for theses standard platforms. The performance of the system is evaluated using experimental data sets acquired for subjects wearing this system during an exercise sequence. This same approach can be used in context-aware monitoring of diverse physiological signals in a patient’s daily life. Furthermore, a new energy-aware wearable system is introduced. It is capable of performing real-time energy profiling on major components through a convenient software interface. Exploring the techniques on how to utilize this energy information and optimize the existing context-aware algorithm is the focus of future work.