Towards efficient feature interaction handling

One of the primary objectives behind the IN is to facilitate rapid provisioning of new services. The phenomena of service interaction is regarded as one of the major obstacles to this objective. In order to assure short time-to-market without compromising service quality, service providers have to implement efficient methods for detecting interactions already in the early phases of service creation. So far, unfortunately, the question of efficiency received very little attention from the research community dealing with service interaction. We present an approach called filtering which may improve the efficiency of interaction detection for IN services. The filtering approach aims at reducing the problem space (i.e. the number of service feature combinations to be considered) prior to an extensive interaction analysis. We motivate the proposed approach, introduce the idea of a service usage model (SUM), and explain its semantics and application. We also discuss formalisation of the SUM with Petri nets. The formalisation is introduced step-by-step on the basis of the service examples already given. Possible extensions of the original SUM are presented. Finally, the pro and cons of the proposed approach as well as some areas for future work are discussed