A QoS preference-based algorithm for service composition in service-oriented network

Quality of service (QoS) based provisioning for service-oriented network applications has been actively discussed because of user's specific requirements. An order relation vector model is proposed in this paper, which can be used to represent and calculate user's preference information effectively. User's preference order of QoS attributes is a possible approach to formulate user's specific QoS requirements in the context. In order to apply the preference order to actual application, a QoS attribute matrix is introduced to manage QoS attribute value; at the same time, a feasible method of defuzzifying and normalizing is also introduced to deal with the original QoS attribute matrix; user's preference order and the normalized QoS attribute matrix are used to calculate user's QoS attribute weight vector, which can be regarded as an evaluation criterion reflecting user's different QoS preference degrees. Finally, a service composition model and algorithm is proposed in this paper. The numerical examples indicate that the model delivers user's QoS preference reliably and the experiment shows that the algorithm is efficient.