A tool for QoS probability distribution estimation of web service compositions

In this paper, we develop a QoS DIstribution Estimation Tool: QoSDIST for service compositions. QoSDIST has the following functions which distinguish it from existing QoS estimation approaches: (1) QoSDIST can generate much more accurate QoS probability distributions for componentWeb services than existing methods [1]; (2) When estimating the QoS probability distribution for a service composition, QoSDIST does not put any constraints on the representation of the QoSs of component Web services, i.e., the QoS of a component Web service can be in single value, discrete values with frequencies, standard statistical distribution, or any general distribution regardless of its shape, which can not be done by any existing approaches; (3) QoSDIST can deal with commonly used composition patterns, including loop with arbitrary exit points.