Swarm-intelligence-based Approach For QoS-aware Service Composition

Web services are recognised as the dominant technology for Server-Oriented Architecture(SOA).They are cross-platformlow-coupling,reusable,reconfigurable and highly interoperable,and these features allow applications with complex functionality to be quickly integrated with multiple individual Web service components to meet users' requirement.With the rapid development of service computing and Web technologies,many enterprises and organisations are providing a large number of Web service components,resulting in a large number of candidate combination solutions for a service composition application.How to select the combination solution that satisfies user's requirements from the huge number of composition schemes is a challenging task.Among swarm intelligence algorithms are a hot research topic for optimising such problems.Swarm intelligence is a class of intelligent groups with self-organising behaviour that evolve complex,highly intelligent behaviour through collaboration between simple individuals,thereby completing the computation of complex problems.And the Harris Hawks Optimization algorithm(HHO)is the latest swarm intelligence algorithm proposed in 2019,which is rapidly gaining popularity because of its simplicity,effectiveness and strong robustness.In this thesis,I construct two frameworks for the combinatorial optimisation of Web services based on HHO algorithm.Firstly,to address the shortcomings of the original HHO in terms of insufficient convergence accuracy and long computation time,an enhanced hybrid algorithm Harris Hawks Optimization algorithm with Elite Evolutionary Strategy(EESHHO)is proposed to improve its convergence speed and accuracy by exploiting the dominant dominant characteristic of elite individuals.A framework for Web service composition in integer coding mode based on the EESHHO is also constructed.Simulation results show that the framework performs better than other mainstream algorithms in most service composition scenarios.However,because of the discrete spatial nature of the service composition problem,the algorithm can easily fall into disorderly search in large-scale data scenarios,leading to poor optimisation search accuracy.With this motivation,our work proposes a pre-processing method for candidate service sets,called fuzzy continuous neighborhood relations.The method can give continuous space properties to the service composition model,thus improving the swarm intelligence algorithm's population search performance.Meanwhile,The Logiest chaotic sequence is introduced into the original HHO for unidimensional perturbation to improve HHO's local ergodicity.This improved algorithm is called CHHO.Also,the CHHO-based framework for Qo S-aware Web service composition is constrcucted.Simulation results show that the framework maintains excellent performance on large-scale datasets.