Personalized Quality Prediction for Dynamic Service Management Based on Invocation Patterns

Recent service management needs, e.g., in the cloud, require services to be managed dynamically. Services might need to be selected or replaced at runtime. For services with similar functionality, one approach is to identify the most suitable services for a user based on an evaluation of the quality (QoS) of these services. In environments like the cloud, further personalisation is also paramount. We propose a personalized QoS prediction method, which considers the impact of the network, server environment and user input. It analyses previous user behaviour and extracts invocation patterns from monitored QoS data through pattern mining to predict QoS based on invocation QoS patterns and user invocation features. Experimental results show that the proposed method can significantly improve the accuracy of the QoS prediction.