Research On Task Resource Matching Algorithm In Centerless Network

As continual renewing of technology and service in modern network, a novel environment of center-less network is emerging. To improve the performce of center-less network, appropriate strategies and algorithms of task-resource matching are of importance. Some of traditional algorithms have been investigated while concurrency of matching for multi-task with flexible multi-QoS constraints of them need to be further researched. Taking grid computing as an example, in this thesis, task-resource matching is studied in center-less network environment and a multi-QoS constrainted algorithm for multi-task simultaneously, TRAM, is proposed in following several aspects.1. Matching the optimal resource considering multi-QoS constraints. After lucubrateing characteristics of center-less network, such as dynamism, heterogeneity, autonomy, etc., multi-QoS constraints and their relations are formulated as MADM problem in TRMA, and extended AHP is used to solve MADM. While considering multi-QoS constraints, the optimal resource is matched for one task with TRMA.2. Matching optimal resources for multiple tasks concurrently. Analyzing differentia of matching resources for tasks orderly or concurrently, matching between tasks and resources is formulated as bi-graph problem in TRMA. Optimal resources can be matched for tasks as possible by Gale-Shapley algorithm, and matched resources will work efficiently.3. Evaluating quality of TRMA. Implementing algorithms of FCFS, Round-Robin, and TRMA separately in simulative environment of center-less network, it is proven that TRMA is scalable in multi-QoS constraints and flexible in scales of tasks and resources.