Dynamic multi-objective service composition based on improved social learning optimization algorithm

Abstract

The technique of multi-objective service composition (MOSC) aims to create powerful and large grained services through aggregating some simple services while optimizing some conflicting objectives, which has become a hot topic in fields of service computing, cloud computing, cloud manufacturing, and so on. Due to the dynamic nature of the services and users’ objectives, the problem of MOSC is actually an dynamic multi-objective service composition problem (DMOSC). Recently, although some wonderful research has been conducted on multi-objective service composition, few studies have paid attention to the issue of DMOSC. To fill this gap, we propose the problem of DMOSC for the first time and then develop a method for DMOSC based on the improved social learning optimization algorithm (named DMOSCO). First, we construct the mathematical model for the problem of DMOSC considering the dynamics of services and users’ objectives. Then, we devise the changes perception strategy and changes response strategy to deal with the changes of candidate services and users’ objectives during the service composition process, which is designed to strike a balance between convergence and diversity within improved SLO. Finally, to solve the problem of DMOSC, we improve the SLO algorithm by designing a new obsvervation learning operator and incorporating the changes detection strategy and response strategy. Extensive experimental results have proved the performance superiority of our proposed method.