Loading...

Optimization of Service Composition Problem in Cloud Manufacturing

Kerdegari, Adeleh | 2015

734 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 47587 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Eshghi, Koroush
  7. Abstract:
  8. Service composition is an important problem in the cloud manufacturing paradigm in which, after receiving a customer request, combination of cloud services are determined for accomplishment of customer needs. Actually, customer need is decomposed to some distinct tasks such that each one is done by a company or a group of them. In this way, virtual and cloud-based production line can be developed based on requirements of customers. The ultimate goal of service composition is optimum assignment of tasks to factories while some objective functions (e.g. minimization of cost and time) and constraints (e.g. minimum reliability and availability) are considered. In this study, integer programming model of service composition problem in four structure of sequence, parallel, loop and hybrid is presented. Then, two different scenarios (in sequence and hybrid structure) of service composition problem are generated such that each center of Iran's provinces includes a factory which can perform some of pre-determined tasks. Also, distance of center of Iran's provinces as well as transportation time between them are estimated based on real data. In order to solving mentioned scenarios, branch and bound exact algorithm and exact optimization software are used. Furthermore, before developing a meta-heuristic algorithm for implementation in large-size service composition problem, landscape analysis of the problem is completed. Based on results of this analysis, the problem has a random uniform nature and its local optima are scattered over the search space. As a result, single-solution based algorithms starting with a random feasible solution can converge to a local optimum after a number of iterations. Hence, a simple single-solution based algorithms such as local search heuristic can search a rugged plain landscape effectively and find a valid solution quickly. Results of comparison between branch and bound and local search algorithms indicate superiority of local search algorithm in finding optimum or near-optimum solution with lower computational cost
  9. Keywords:
  10. Cloud Computing ; Optimization ; Cloud Manufacturing ; Service Composition ; Local Search ; Landscape Analysis

 Digital Object List

 Bookmark

No TOC