Implementation of virtual sensors in body sensor networks with the SPINE framework

Signal processing for body sensor networks usually comprises multiple levels of data abstraction, from raw sensor data to data calculated from processing steps such as feature extraction and classification. This paper presents a multi-layer task model based on the concept of virtual sensors (VS) to improve architecture modularity and design reusability. In our pilot application of gait parameter extraction, VS are abstractions of components of BSN classification systems that include sensor sampling and processing tasks and provide data upon external requests analogous to the function of physical sensors. The paper presents an extension of the SPINE framework including a new buffer management scheme that facilitates the VS implementation.