Feature interactions detection in intelligent networks

Intelligent Networks (IN) have been introduced for rapid development and deployment of new services. However, these new services may interact with old ones in a negative and unexpected manner. This is known as feature interaction (FI) problem. In this thesis, a new pragmatic approach for feature interaction detection is proposed. A pragmatic method for FI detection is based on the gained experience from the study of the known FIs. In this approach, the causes of FIs given in Bellcore and European benchmarks serve as the starting point. The sufficient information for feature description is derived from these causes. This information is modelled using an object-oriented template. Features are described in terms of necessary resources and actions. Feature participants call model, their triggering constraints and operations describe the behaviour of the feature. The detection method takes these models as input and checks for interactions between features. The method consists of three parts: filtering algorithm, feature instantiation and detection algorithm. Using the filtering algorithm, all the possible interaction-prone scenarios are produced and the actual features are instantiated using the list of topologically different call combinations. The detection algorithm is run on these actual participants of the features in order to detect potential interactions. The approach has been implemented as a tool and applied successfully to the existing feature interaction benchmarks