Energy-efficient context classification with dynamic sensor control.

Energy efficiency has been a longstanding design challenge for wearable sensor systems. It is especially crucial in continuous subject state monitoring due to the ongoing need for compact sizes and better sensors. This paper presents an energy-efficient classification algorithm, based on partially observable Markov decision process (POMDP). In every time step, POMDP dynamically selects sensors for classification via a sensor selection policy. The sensor selection problem is formalized as an optimization problem, where the objective is to minimize misclassification cost given some energy budget. State transitions are modeled as a hidden Markov model (HMM), and the corresponding sensor selection policy is represented using a finite-state controller (FSC). To evaluate this framework, sensor data were collected from multiple subjects in their free-living conditions. Relative accuracies and energy reductions from the proposed method are compared against naïve Bayes (always-on) and simple random strategies to validate the relative performance of the algorithm. When the objective is to maintain the same classification accuracy, significant energy reduction is achieved.

CARER: efficient dynamic sensing for continuous activity monitoring.

Advancement in wireless health sensor systems has triggered rapidly expanding research in continuous activity monitoring for chronic disease management or promotion and assessment of physical rehabilitation. Wireless motion sensing is increasingly important in treatments where remote collection of sensor measurements can provide an in-field objective evaluation of physical activity patterns. The well-known challenge of limited operating lifetime of energy-constrained wireless health sensor systems continues to present a primary limitation for these applications. This paper introduces CARER, a software system that supports a novel algorithm that exploits knowledge of context and dynamically schedules sensor measurement episodes within an energy consumption budget while ensuring classification accuracy. The sensor selection algorithm in the CARER system is based on Partially Observable Markov Decision Process (POMDP). The parameters for the POMDP algorithm can be obtained through standard maximum likelihood estimation. Sensor data are also collected from multiple locations of the subjects body, providing estimation of an individual's daily activity patterns.

Episodic sampling: towards energy-efficient patient monitoring with wearable sensors.

Energy efficiency presents a critical design challenge in wireless, wearable sensor technology, mainly because of the associated diagnostic objectives required in each monitoring application. In order to maximize the operating lifetime during real-life monitoring and maintain sufficient classification accuracy, the wearable sensors require hardware support that allows dynamic power control on the sensors and wireless interfaces as well as monitoring algorithms to control these components intelligently. This paper introduces a context-aware sensing technique known as episodic sampling - a method of performing context classification only at specific time instances. Based on Additive-Increase/Multiplicative-Decrease (AIMD), episodic sampling demonstrates an energy reduction of 85 percent with a loss of only 5 percent in classification accuracy in our experiment.

MEDIC: medical embedded device for individualized care.

OBJECTIVE: Presented work highlights the development and initial validation of a medical embedded device for individualized care (MEDIC), which is based on a novel software architecture, enabling sensor management and disease prediction capabilities, and commercially available microelectronic components, sensors and conventional personal digital assistant (PDA) (or a cell phone). METHODS AND MATERIALS: In this paper, we present a general architecture for a wearable sensor system that can be customized to an individual patient's needs. This architecture is based on embedded artificial intelligence that permits autonomous operation, sensor management and inference, and may be applied to a general purpose wearable medical diagnostics. RESULTS: A prototype of the system has been developed based on a standard PDA and wireless sensor nodes equipped with commercially available Bluetooth radio components, permitting real-time streaming of high-bandwidth data from various physiological and contextual sensors. We also present the results of abnormal gait diagnosis using the complete system from our evaluation, and illustrate how the wearable system and its operation can be remotely configured and managed by either enterprise systems or medical personnel at centralized locations. CONCLUSION: By using commercially available hardware components and software architecture presented in this paper, the MEDIC system can be rapidly configured, providing medical researchers with broadband sensor data from remote patients and platform access to best adapt operation for diagnostic operation objectives.

Context-aware sensing of physiological signals.

Recent advancement in microsensor technology permits miniaturization of conventional physiological sensors. Combined with low-power, energy-aware embedded systems and low power wireless interfaces, these sensors now enable patient monitoring in home and workplace environments in addition to the clinic. Low energy operation is critical for meeting typical long operating lifetime requirements. Some of these physiological sensors, such as electrocardiographs (ECG), introduce large energy demand because of the need for high sampling rate and resolution, and also introduce limitations due to reduced user wearability. In this paper, we show how context-aware sensing can provide the required monitoring capability while eliminating the need for energy-intensive continuous ECG signal acquisition. 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 these 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.