Doctor of Philosophy
School of Information Systems and Technology
Wang, Lijuan, Bio-inspired cost-aware optimization for data-intensive service provision, Doctor of Philosophy thesis, School of Information Systems and Technology, University of Wollongong, 2014. http://ro.uow.edu.au/theses/4241
The rapid proliferation of enormous sources of digital data and the development of cloud computing have led to greater dependence on data-intensive services. Each service may actually request or create a large amount of data sets. The scope, number, and complexity of data-intensive services are all set to soar in the future. To compose these services will be more challenging. Issues of autonomy, scalability, adaptability, and robustness, become difficult to resolve. Bio-inspired algorithms can overcome the new challenging requirements of data-intensive service provision. It is useful for the provision of data-intensive services to explore key features and mechanisms of biological systems and accordingly to add biological mechanisms to services.
This thesis discusses the key concepts and approaches behind Web service concretization. It presents a hierarchical taxonomy of Web service concretization approaches and provides a detailed analysis of each approach. The taxonomy helps us not only to identify existing algorithms developed in service concretization but also to evaluate the applicability of different approaches to the data-intensive service provision problem.
This thesis concentrates on the bio-inspired algorithms in the taxonomy. To this end, it conducts a systematic review of Web service concretization based on bio-inspired algorithms. The systematic review investigates the extent of applications of bio-inspired algorithms to QoS-based Web service concretization, which is not covered by previous studies. Then the comprehensive applications of an ant colony system and a genetic algorithm for cost-aware data-intensive service provision are introduced. The performance of the two proposed algorithms is evaluated and compared with other traditional methods. Further, this thesis investigates an ant colony system and a genetic algorithm for the multi-objective data-intensive service provision. Both the algorithms get a set of Pareto-optimal solutions by considering the total cost and the total execution time of a composite service at the same time. Experiments using many different scenarios with respect to five performance metrics have been conducted to evaluate the proposed algorithms.
This thesis then further studies an ant colony system for the dynamic dataintensive service provision problem. In the dynamic environment, the service composition optimization process should be conducted repeatedly when the changes of the states of services occur. In order to adapt the ant colony system to handle the dynamic scenarios, several pheromone modification strategies in reaction to changes are discussed. The aim of the strategies is to find a balance between preserving enough old pheromone information to speed up the search process, and resetting enough new pheromone information to facilitate the ants to find a good solution for the changed scenarios. The strategies differ in their degree of reinitialized pheromone values with respect to the information that has been used to decide the amount of pheromone values. Moreover, the behavior of different strategies for modifying pheromone information is compared.
Data-intensive services are typically used in a dynamic and changing environment, and different providers typically have conflicting objectives. In order to automate the process of reaching an agreement in the data-intensive service provision, this thesis finally proposes an ant-inspired negotiation approach. It uses a group of agents automatically negotiating to establish agreeable service contracts. The lifetime of the complete data-intensive service provision and the two-stage negotiation procedures are described. A multi-phase, multi-party negotiation protocol is also designed. The negotiation approach is evaluated through simulations.
This thesis also points out future research topics when it concludes its contributions.