Estimation of Unknown Quality of Service Values to Increase the Quality of Composite Grid Service Using Bayesian Network

Computational Grids have developed as a new approach to solve large scale problems in scientific، engineering and business areas. Open Grid services architecture is an adaptation of the service-oriented architecture which presents the Grid operation a set of service-oriented softwares. Grid services composition provides the possibility for users to submit their complex requirements as a single request. QoS-aware Grid service composition algorithms try to construct a composite Grid service which satisfies the user-defined constraints as well as to optimize the QoS parameters. All of the presented service composition approaches in the literature discard using the Grid services with unknown QoS values in the composition process. However، estimation of unknown QoS values of Grid services provides an opportunity for them to be selected as component Grid services in the constructed composite service. In this paper، a probabilistic QoS model is presented for estimating the unknown QoS values using Bayesian network. Experimental results indicate that estimating the unknown QoS values has high accuracy and leads to more efficient composite Grid services.