Heart rate monitoring and user behavior anomaly detection using DynaSense
MetadataShow full item record
There are a number of user centric applications that use data from sensors in the personal area network. The heavy dependence of such apps on sensors means that if the user forgets to carry the sensor device, the apps fail. However, the data generated from a given sensor that is unavailable can be derived from other devices or combination of sensors. Since it is not possible for app developers to track all such scenarios and to interface with a given sensor embedded in all possible devices that a user can have, user apps cannot take advantage of the sensor rich environment of prospective users. This thesis introduces the design of a middleware that allows user apps to be agnostic of the data source or sensor in use. The middleware allows data source driver apps to define a tree like structure of data sources to derive new data sources which can be delivered to all requesting user apps. The middleware dynamically decides which sensors to use for a given data source and returns data to user apps using a callback mechanism. To compare how user applications can be composed with fewer lines of code and higher code reuse, an application for detecting anomalies in a user's daily behavior and a heart rate monitor are implemented using the middleware.