Constrained Resources for Mobile Health Monitoring and Remote E nvi ron m en tal Mon i to ri ng

Recent advancements in sensors, wireless technology, and a reduction in the form factor of computing devices, have pushed us closer to the realization of true autonomy in mobile sensing systems. Past fielddeployable sensing systems for healthhiomedical applications and even environmental sensing have been designed for data collection and data transmission at pre-set intervals, rather than for on-board processing. This lack of true autonomy has resulted in systems with lower lifetimes and those that require large amounts of bandwidth to transmit all sensory data at all times. We propose the use of a neM: general machine learning architecture that can be used for a variety of autonomous sensing applications that have very limited computing, power, and bandwidth resources. We lay out general solutions for eflcient processing in a multitiered (three-tier) machine learning framework that is suited for remote, mobile sensing systems with low computing capabilities. Simple pattern recognition methods are used at the sensor level to filter signijkant events. Novel dimensionality reduction techniques that are designed for classijkation are used to compress each individual sensor data and pass only relevant information to the mobile multisensor fusion module (second-tier). Statistical classijers that are capable of handling missing/purtial sensory data due to sensor failure or power loss are used to detect critical events and pass the information to the third tier (central server) for manual analysis and/or analysis bj’ advancedpattern recognition techniques. The applicability of our technoloa in mobile health & alcohol monitoring is shown. Other uses of our solution are also discussed.