Experiential learning-based feature interaction detection in IMS

The detection of feature interaction in IMS remains a challenging task. This paper investigates the feature interaction induced by multi-user and multi-service, analyses and describes them through formal language, and proposes an experiential learning-based detection algorithm. This algorithm regards the interaction occurring for the first time as an experience, which is self-learned from an actual conflict, when it recurs, the previous detection can provide some experience for the later detecting process, to predict interaction in advance. So the detection time can be advanced as well as the resolution time is reduced. Case study shows that this approach is effective and can detect the feature interaction much earlier than traditional methods.