Since users are surrounded with various devices in the dynamic and constantly changing ubiquitous computing environment, users need more adaptive and personalized application services. However, it becomes more difficult to realize users’ needs due to the ever increasing complexity of the task and the limitation of the capability of a single service. Creation of a new service on-demand to carry out more complex tasks is possible through the composition of existing services. As a result, there have been a lot of interests in defining and implementing new mechanisms for service composition in industry and academia to address challenging problems. Since ubiquitous computing environments are very dynamic and unpredictable, Artificial Intelligence planning methods are more suitable and useful to create dynamic service composition than other methods. Planning is one of the most promising techniques for the automated composition of services.
In this thesis, we present a developed a goal-based approach for service composition using HTN planning method. For the easiest way of express user requirement, we consider an abstract goal description using ontology. In our system, a user requests an abstract goal in an intuitive manner and composes the requested service based on service ontology. User request is refined into primitive goal through goal refinement using semantics of ontology. Each primitive goal is an input of planner, which decompose an input into primitive task. Then, we can make a new composite service which composed of goal graph (from top-goal to sub-goals) and task graph (from top task to subtasks). The proposed architecture consists of a service request analyzer, HTN planner, service composer, and service validator. Service developer can create a new service using planning-based service composition description. We have considered a case study in the smart office domain and energy aware system domain for proving its suitability to the objectives of our service composition system.
To evaluate the proposed architecture, we have carried out experiments for case studies. Experimental results show that our proposed model and architecture is good for adaptive service composition.