Skip to main content
eScholarship
Open Access Publications from the University of California

Quality of Information Driven Environment Crowdsourcing and its Impact on Personal Wellness Applications

  • Author(s): Matthews, Jerrid E.
  • Advisor(s): Gerla, Mario
  • et al.
Abstract

Mobile devices with programmable embedded sensors and internet access have enabled a new paradigm of socially beneficial software applications. These devices may be stationary or mobile, and located sparsely across the globe operating under heterogeneous environments. These multi-lateral sensor data feeds produced by both autonomous and human sensing agents can be aggregated and transformed by a system to produce a human understandable spatiotemporal representation of a phenomenon (ie: event) in real-time. These data can then be disseminated using many different communication infrastructures (e.g. 4GLTE, WiFi). The study of how to efficiently organize these complex sensor data feeds is the primary contribution of this dissertation; in addition we present two health and wellness sensor data applications that leverage the sensor data feeds. Traditional sensor data platforms require the data publisher to associate a set of descriptive terms (also known as keywords or tags) with their data feed in order to organize the sensor data. Information operators must perform a keyword-based search in order to retrieve the data feeds of interest, which may require subject matter expertise to identify relevant keywords. The central theme of this thesis is the leveraging of personal sensor platforms, internet computing resources and crowdsourcing campaigns to achieve not only individual wellness but also community health maintenance. We contribute a new ontological data model for organizing and enriching sensor data with valuable QoI/VoI attributes. In addition, we combine theoretical models and systematic measurements to show that it is possible to organize sensor data in such a way to retrieve relevant sensor data in order to measure a phenomenon of interest without tagging or human input.

Main Content
Current View