PLR: A benchmark for probabilistic data stream management systems

Abstract

Inherent imprecision of data streams in many applications leads to need for real-time uncertainty management. The new emerging Probabilistic Data Stream Management Systems (PDSMSs) are being developed to handle uncertainties of data streams in real-time. Many approaches have been proposed so far but there is no way to compare them regarding precision and efficiency. This problem motivated us to design an evaluation framework to compare performance and accuracy of PDSMSs with each other and also with probabilistic databases. In this paper, after a brief introduction to PDSMSs, we describe requirements and challenges for designing a PDSMS benchmark. Then, we present different parts of our framework including probabilistic data stream generator, queries, and result evaluator. Furthermore, we focus on implementation aspects and use our framework to evaluate effects of floating precision in our PDSMS prototype. © 2012 Springer-Verlag.