Semantic-based service analysis and optimization

Abstract

The need to autonomically optimize end user service experience in near real time has been identified in the literature in recent years. Service management systems that monitor end user service session context are deployed but approaches that estimate end user service experience from session context do not analyse the compliance of that experience with user expectations. Approaches that plan and execute actions to optimize end user service session delivery are not applicable to arbitrary service sessions; they work with specific service types and delivery mechanisms or do not consider end user service experience when making optimization decisions. Another barrier to autonomic end user service management optimization is the lack of a holistic model for the domain. This thesis proposes the Aesop approach, an approach that addresses semanticbased autonomic optimization of end user service delivery. This approach has a knowledge base at its core and proposes the EUSAO ontology. This ontology, designed to semantically model the end user service management domain, enables partitioning of knowledge that varies over time for efficient access. The Aesop Engine is designed to execute an iteration of an autonomic loop (Monitor, Analyse, Plan, Execute) in near real time. It runs semantic algorithms designed to use queries and rules on subsets of the partitioned EUSAO-based knowledge in order to monitor end user sessions, to analyse their compliance with expectations, to plan optimizations, and to execute those optimizations as throttling actions on the service delivery network. The semantic-based algorithms that are proposed are efficient because they operate on small partitioned subsets of the knowledge base. The Aesop approach allows arbitrary end user service types and network domains to be added by specializing the EUSAO ontology for that domain and adding domain specific semantic mappings, queries and rules. A case study has demonstrated that the approach is applicable in the Mobile Broadband access domain. A prototype implementation of the complete Aesop approach was evaluated on a full purpose built Home Area Network test bed, on which execution of end user service sessions was automated and controlled. In the evaluation, when the measured compliance of a set of end user service sessions with expectations when Aesop optimization was active was compared with the compliance of an identical set of sessions when Aesop optimization was inactive, a significant improvement was observed on the compliance levels of high priority sessions in all experimental scenarios, with compliance levels more than doubled in some cases. The key contribution of the Aesop approach is that it enables user centric service delivery management. When used with other service specific and network centric approaches, it shows promise as a method of optimizing an arbitrary running set of end user service sessions so that they best meet the expectations of the individual users running those service sessions. In addition, this research has addressed key concerns that have inhibited the widespread adoption of semantic techniques by showing that, with appropriate design and implementation, semantic techniques can be efficient and can perform well, and are deployable in full autonomic systems that execute in near real time.