Our project is motivated by the following monitoring: applications:

  1. Object tracking and monitoring with RFID technology:

    RFID technology has been widely adopted in many applications such as library, supply chain and retail management. Data streams from RFID readers are highly noisy due to the sensitivity of sensing of the reading and other environmental factors such as metal objects and interference. The raw data only contains tag identities and doesn't reveal high-level information such as object location information.

  2. Weather monitoring:
  3. Meteorological data streams are collected from a weather radar network and processed in a real-time stream system to predict natural disasters such as tornados. Uncertainty in such data arises from environmental noise, device noise and inaccuracies of various radar components. The fact that data collected from these radar is high-volume requires efficient processing to meet stream speed.

  4. Vehicle tracking, Linear Road benchmark:

    A highway/road system can be equiped with a sensor network to monitor traffic condition for further actions. Data streams are noisy due to device and environmental noise.

  5. Computational Astrophysics:

    Advanced telescopes are employed in the sky survey to measure the positions and properties of all the hundreds of millions of celestial objects. Most attributes that resulted from scientific measurements or their data cooking processes are uncertain. As a concrete example, the Galaxy table in the SDSS astronomy archive has 297 attributes, out of which 151 attributes are uncertain (click here to see the schema).

Last Update: August 2016