A Hybrid Service Selection Approach for Multi-user Requests

Present web service selection methods are trying to find solutions that can get an optimal QoS utility value for single user while satisfying user´s requests. However, affections from other users´ solution are lack of consideration, especially for those users who have same function requirements but with different QoS preference. From the perspective of network operators, they wish to find solutions for all users that can get an optimal QoS utility value, while meets each individual requirements and global constrains. Besides, as users grow, computation time for global optimal solution using traditional method will be intolerable. Therefore this paper proposes a hybrid method combining Genetic Algorithm with Ant Colony Algorithm and KM algorithm to meet above all requirements. Firstly, we decompose global QoS constrains into local constrains in dimension of service class. Secondly, decomposition of each service class´s local constraint is taken in user dimension, transforming global selection problem into local selection problem. Thirdly, results got by above two steps are used to achieve the final overall-optimal solution. Experiments show that our hybrid method is particularly suitable for situation of large-scale multi-user service selection and outperforms traditional solutions in terms of computation time while achieving close-to-optimal results.