A flexible QoS-aware Web service composition method by multi-objective optimization in cloud manufacturing

Highlights

• The QoS-aware service composition is optimized based on multi-objective optimization.

• Users gain decision flexibility with tradeoff between QoS performance and QoS variance from their expectation.

• An efficient EDMOGA is developed to achieve outstanding convergence and diversity of the Pareto frontier.

Abstract

Cloud manufacturing, combining Web services via internet to a cooperative manufacturing system, has been an increasing popularity for global manufacturing. It will unlock the tremendous value in the massive amount of data being generated by the manufactories. The problem of QoS-aware Web service composition (QWSC), i.e., selecting appropriate service for each component of a service composition from a pool of functionally identical service to satisfy the users’ end-to-end QoS constraints, is a core of the cloud manufacturing. A novel QWSC method by multi-objective optimization is proposed to help users to make a flexible decision. First of all, the problem of QWSC is formulated to a multi-objective optimization model where either QoS performance or QoS risk (variance comparing to the user‘s QoS requirement) is the individual optimization objective. And then, an efficient ε-dominance multi-objective evolutionary algorithm (EDMOEA) is developed to solve the presented model. Finally, experimental results verify the effectiveness and efficiency of the proposed method for the large-scale QWSC problem.

Introduction

Nowadays, cloud computing or service-oriented computing (SOC), where Web service is the essential element to support rapid, low-cost, and easy composition of distributed applications in heterogeneous environments; have become increasingly popular, and triggers a considerable amount of research efforts in both academia and industry (Papazoglou, Traverso, Dustdar, & Leymann, 2007). SOC provides a new opportunity to the modern manufacturing industry to do e-manufacturing and business collaborations for global manufacturing, which aims to align business processes across the organizations (Wu, Greer, Rosen, et al., 2013). Based on the trends in devices, growth of data, communication bandwidth, and growth of peer to peer sharing, it is argued that utilizing Web services offered by third-parties with a greatly reduced cost is emerging for manufacturing industries that will unlock the tremendous value in the massive amount of data being generated by the industrial manufactories (Tao & Zhang, 2012).

In most cases, individual Web services are combined to a composite service to solve a complex goal successfully. Fig. 1 shows an example of composite service used in warehouse fire alarm, where individual devices providing simple function can be combined to establish a more powerful alarm service. It involves different tasks including temperature sensors, smoke detectors, infrared sensors, liquid immersion sensors, alarm bells, camera, scupper valves and water sprinklers. Each task can be bound to various Web services providing the same functionality but with different quality properties.

As many Web services deliver the equivalent functionality, Quality of Service (QoS), such as the cost, response time, reliability, and availability are crucial concern to help users to select appropriate services satisfying their preferences, for instance, cost should be less than $20, and response time should be less than two hours. The concern with QoS introduces a challenging problem of QoS-aware Web service composition (QWSC), i.e., select the best composite service with optimal QoS performance while satisfying given end-to-end QoS constraints (Qi, Dou, Zhang, & Chen, 2012). Unfortunately, due to the increasing number of component services (tasks) in a composite service and the growing spread of candidates for each task, the size of possible composition plans built by selecting different Web service instance for each component service should be in general exponentially enlarging. In other words, the QWSC is a NP-Complete problem in a strong sense which cannot be solved within a reasonable time (Haddad, Manouvrier, & Rukoz, 2010).

Motivated by the large-scale QWSC problem, this study propose a flexible and efficient Web service selection method by multi-objective optimization to provide users enhanced information to choose appropriate services. A QoS model for describing vague concept of QoS with specific features is constructed. Furthermore, traditional optimization methods for QWSC tend to find the best composite service fulfilling user’s overall QoS constraint. They may fail to provide a possible solution if unsuitable QoS constraints are submitted. Unfortunately, it is impossible that users are always well-informed to submit reasonable QoS requirements. From this perspective, then the problem of QWSC is formulated into a multi-objective optimization model where either QoS performance or variance from the user’s QoS constraints (i.e. QoS risk) is an independent objective. Moreover, a ε-dominance multi-objective evolutionary algorithm (EDMOEA) is developed to solve the presented model. Finally, experimental studies demonstrate the effectiveness and the efficiency of the proposed method.

The remainder of this paper is organized as follows. Section 2 gives an overview of related work. Section 3 presents a QoS model of Web service that defines the QoS criteria including cost, execution time, latency time, reliability and availability; and the QoS aggregation for a composite service. The multi-objective optimization model for QWSC and the solving algorithm are specified in Sections 4 Multi-objective optimization model for service selection, 5 Multi-objective optimization solving algorithm design respectively, Section 6 presents the experimental studies, while Section 7 offers the concluding remarks and outlines future study.