Toward Ubiquitous Environment: An Scalable Framework for Autonomous Service Composition

In ubiquitous environment, there are various services provided by different underlying devices, which need cooperate together with not only the correct functional workflow, but also the non-functional composition quality. In this paper, we propose an integrated and scalable framework for quality of services (QoS) aware autonomous service composition. In addition to QoS, we also consider the execution cost in the composition which can be reflected on the number of services. It consists of three stages: 1) construction of the dependency graph that descripts the relations among corresponding services for a specific request; 2) QoS aware graph pruning, is divided into strict strategy and loose strategy; 3) composition solution search, including greedy near-optimal search based on strict pruning strategy and global optimal search based on loose pruning strategy. Experimental results prove that our framework outperforms other selected algorithms or frameworks in composition effect and processing efficiency, especially when processing large scale services.