Research On Composite Service Skyline With QoS Correlations

In the service-oriented architecture, resources distributed in the Internet are en-capsulated as publically-accessible Web Services, which can be further orchestrated to make composite services with well-defined processes (also known as workflows). This process is called web service composition. The key to web service composition is how to efficiently select services from a large number of similar functionality services with satisfying the consumers’ requirements.Composite service broker constructs a composite service which satisfies a con-sumer’s preferences and constraints of QoS. But once the preferences change, the composite service broker needs to reselect the optimal one for the user by taking a comparison in the whole candidate services which is inefficiency. Composite service skyline is a set of composite services which are not dominated by others. Composite service skyline is usually used to improve the efficiency of composite service selection. When the weights of user’s preferences change, the broker just need to traverse a small space to get the optimal one. State-of-the-art approach to computing composite service skyline assumes that candidate services of different tasks are independent, and the ser-vices are running in the non-mobile devices. But QoS correlations between services are commonly existing in practical applications. In addition, along with the spread of mobile network, and the appearance of smart autonomous devices, the runtime envi-ronment of web service tend to diversification:not only be a cloud service, but also a smart autonomous device. Because of the mobility of devices, it is clearly that services provided by the mobile devices have the feature of mobility. The composite service skyline may change, because QoS or correlated quality values of a service may change as mobile devices move.These issues cause a lot of new challenges to composite service skyline com-puting. In this paper, we investigate these issues respectively, and provide a solution for each of them. Specifically, we propose a service model supports QoS correlation, and based on this model, we put forward a method to compute composite service sky-line. We present several pruning criteria based on which two pruning algorithms are proposed to accelerate our approach. Then a new concept of safe value range is pre-sented aim to improve the efficiency of computing composite service skyline in mobile environment, we also propose two methods to compute and update safe value range respectively. At last, we conduct a series of experiments to evaluate the effectiveness and efficiency of our algorithms.