Transactional and QoS-aware dynamic service composition based on ant colony optimization

Service composition facilitates seamless and flexible integration of applications from different providers. With the growing number of services that offer the same functionality but differ in non-functional properties published online, an efficient approach for dynamic service selection and composition is required. Traditionally, the problem is mostly addressed either from the quality of service (QoS) aspect or from the transaction aspect. In this paper, we first investigate the transactional properties of services and focus on how to compose individual services in a transactional manner, and then formulate the problem of transactional and QoS-aware dynamic service composition. By modeling the problem as a constrained directed acyclic graph, the ant colony optimization algorithm is utilized to seek a near-to-optimal solution efficiently. At last empirical studies are conducted and the experiments show that the proposed approach can approximate the optimal solution well while staying efficient. (c) 2012 Elsevier B.V. All rights reserved.