Research On Key Technologies For Multi-User Service Selection On Converged Network

As network convergence evolves, the ubiquitous of converged network makes the provision of personalized services in a much convenient and popular way, and more and more user requirements are raised on the network. Thanks to the rapid development of web service technology, a large number of services which are functional similar but QoS different are emerging on converged network. Service candidates with different functionalities can be combined into a composite service workflow to satisfy user’s complex requirements. How to select a feasible single service candidate or composite web service to meet the demands of different users both functionally and non-functionally has become an important issue in area of web service selection.Since existing web service selection researches are mostly focusing on how to find the optimal solution for single user, it is still a challenge to make service selection solutions for multi-user requests while satisfying each user’s personal preference. Besides, the dynamic feature of converged network will lead to the changing of service execution environment. It is very necessary to solve the issue that how to ensure the continuity of service providing and executing based on functional adjustments, if non-functional service selection technology fails to find proper service candidates to satisfy user’s requirements. Although there are some achievements in this research area, it still has following issues for researches of multi-user service selection on converged network.1) The existing QoS normalization approach is too simple. Since different QoS attributes have various meanings and value ranges, simple QoS normalization approach cannot reflect the changing conditions of different QoS attributes.2) The existing service selection criteria are mostly based on QoS utility value, although users’personal preferences are considered as a weight factor in computation process. However, users have different satisfaction value even for the same service candidate and the assumption that users’satisfaction of a service is proportionate to the QoS of service is not always true. Therefore selection conflicts may happen when too many users select the same outstanding web service because present researchers take service’s QoS utility as a judging criterion. Besides, QoS of service may decline due to overload users’requests, since its service ability is limited.3) It is necessary to take adaptations based on functionality to make sure the continuity of service providing when non-functionality selection strategy is unable to find a suitable alternative service candidate. Adaptation description model can be used for analyzing adjustment behaviors and providing guidance for functional-based service selection. However, existing researches are lacking of effective model which can achieve the above requirements.In order to solve these above problems, this thesis takes some deep researches on problem of multi-user service selection on converged network, and makes several significant contributions that are generalized as follows.1) Classify QoS attributes into four types based on QoS attributes’ meaning and value ranges, and present the normalization method according to the QoS type which fits in with QoS changing condition in real world. Then propose a novel user satisfaction calculation method based on the prospect theory of economics. The novel user satisfaction calculation method takes both objective QoS attributes of service and users’subjective requirements into consideration which is good at reflecting the fluctuation of user feeling in real life. In addition, it also considers the factor of users’priority in the calculation. The effectiveness and correctness of this method are proved by the experimental results in simulative environment.2) Propose a hybrid method for multi-user composite service selection problem, and avoid selection conflict effectively. From the perspective of network operator, the hybrid method provides the best fitted services solution for each user according to his or her personal preferences, with the aim of optimizing the whole network performance. The proposed method first decomposes the global constraints into local constraints by using quality level partition twice, and reduces the complexity of the problem gradually. Then it combines ant colony algorithm with genetic algorithm to accelerate the calculation process of composite service selection and introduces non-linear mutation operator to enlarge searching space. Experimental results show that for large-scale users and services, our hybrid method outperforms traditional solutions in terms of computation time while achieving close-to-optimal results.3) Put forward a novel session service adaptation description model. According to the adaptation of telecommunication services, the model analyzes and describes the adaptive adjustments of session service of telecommunication area by multi-angle and multi-layer method. When service selection strategy based on non-functionality cannot find proper alternative service candidate to ensure the continuity of communication process, the model helps to find services that need adjusting based on functional property and provides guidance for further service selection. Comparing with other existing models, the novel model is more appropriate to describe and analyze the adjustment behaviors of adaptable telecommunication service in future converged network.