General probability distribution-based QoS analysis for web service composition

In this paper, we propose a comprehensive solution for QoS aggregation in service composition. Different from existing work on QoS modeling which uses single values, discrete values with frequencies, or well known statistical distributions, the proposed method can work on QoS represented by general probability distributions. A set of formulas are developed to calculate the QoS probability distribution for the composite service based on the four identified patterns. Experimental results are provided to demonstrate the effectiveness of the proposed method.