Comparison of bio-inspired algorithms for peer selection in services composition
RIS ID
31654
Abstract
One of the challenges for the P2P-based service composition process is how to effectively discover and select the most appropriate peers to execute the service applications when considering multiple properties of the requested services. Different ontology-based e-service profiles have been proposed to facilitate handling multiple properties and to enhance the service oriented process in order to achieve the total or partial automation of service discovery, selection and composition. This paper investigates how the ACO (Ant Colony Optimisation) algorithm and the GA (Genetic Algorithm) may facilitate P2P-based (Peer-to-Peer) service selection with multiple service properties. The performance of both algorithms is evaluated and compared statistically using a pooled t-test for 30 randomly generated composition scenarios. Our experimental results show that both algorithms can improve the quality of service composition, while showing that the ACO approach is the more effective.
Publication Details
Shen, J., Beydoun, G., Yuan, S. & Low, G. (2011). Comparison of bio-inspired algorithms for peer selection in services composition. IEEE International Conference on Services Computing (SCC 2011) (pp. 250-257). USA: IEEE.