A Data Analysis Driven Streaming Framework for Body Sensor Area Networks

Streaming and analysis of time series body sensor data have been well investigated recently for various applications, especially on health data monitoring. However, existing strategies work independently from each other. Obviously, lacking appropriate information sharing, feedback, and interaction mechanisms, these strategies, even in combination, do not provide an efficient and effective solution for real time body sensor data collection, transmission, and analysis. In this work, we propose a data analysis driven framework with feedback for efficient streaming of body sensor data. The core idea of this framework is based on a data analysis algorithm specific variance threshold that identifies the data reliability requirement. Then, a reliability index is generated and sent from the data aggregator to sensors as a feedback to guide the streaming protocol. At the sensor side, a data importance ranking and grouping strategy is designed so that samples that affect data analysis most significantly are given higher priority for transmission.