Quality-driven evaluation of trigger conditions on streaming time series

摘要: For many applications, it is important to evaluate trigger conditions on time series streams. In a resource constrained environment, users' needs should ultimately decide how the evaluation system balances competing factors such as speed, result precision, and load shedding level. This paper presents basic framework for algorithms that takes user-specified quality requirements into consideration. Three optimization algorithms, each under different set of requirements, are developed in framework: (1) minimize response given accuracy without shedding; (2) limit requirements; (3) one type errors shedding. Experiments show these effectively achieve their goals while satisfying corresponding requirements.