Sensor data streams exhibit special characteristics such as inherent information uncertainty and inherent data sample correlations, both within and across streams. We introduce a new data model, called Probabilistic Stream Relational Algebra (PSRA), that models a sensor data stream as a set of probabilistic data samples, along with prediction strategies for each attributes, capturing domain knowledge of inherent data correlations. We also explicitly associate every operation with schedule, specifying when next data sample should be produced, to facilitate resource management in sensor networks. We prove that operators in PSRA are non-blocking, thus making PSRA especially suitable for data stream processing. We also show that conventional relational model and existing deterministic data stream processing model can be modeled in PSRA.
Liu, Haiyang; Hwang, San-Yih; Srivastava, Jaideep.
Probabilistic Stream Relational Algebra: A Data Model for Sensor Data Streams.
Retrieved from the University of Minnesota Digital Conservancy,
Content distributed via the University of Minnesota's Digital Conservancy may be subject to additional license and use restrictions applied by the depositor.