Loading...

A Novel Blockchain-Based Optimization Model based on ADMM Method for Cloud Manufacturing Service Composition Problem

Jabbari Marand, Behnam | 2021

407 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54034 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Hoshmand, Mahmoud; Fatahi Valilai, Omid
  7. Abstract:
  8. With the growing product diversification and customization of demands, employing the concept of shared economy and resource sharing is more than ever needed. In this regard, with the advent of the Industry 4.0, IoT , cloud computing, and systems such as cloud manufacturing, impediments have been eliminated to form a network of businesses. Therefore, the problem of service composition, which seeks to allocate the best combination of services to a specific demand, is considered. Due to many services and various quality parameters, these problems fall into the category of mega-size problems. Moreover, in terms of computational complexity, they are from the NP-hard class. Given the service-oriented nature of these problems, which necessitates the real-time achievement of an optimal solution, a decentralized strategy should be adopted. In the present study, a scheduling problem is re-modeled and after being decomposed into subproblems, it is solved using the ADMM method and decentralized platform, namely blockchain. The calculations prove that this technique has reduced the solving time in a scenario by 98.8% compared to the basic ADMM algorithm. In addition, it solves the problem 41.4% faster than the centralized strategy which deals with subproblems sequentially. The performance of the algorithm culminates in scenarios with larger dimensions in which the solving time is much more decreased than smaller ones. Furthermore, the optimality gap using this algorithm is much less than the basic algorithm and studies show that determining smaller values for the relative and absolute errors, leads to an exact optimal solution. However, the basic algorithm takes too long to converge while those values are selected. Unlike other methods, this algorithm not only does not lose its efficiency with the growth of the dimensions of the problem but with more decomposition and using the decentralized strategy, a good answer can always be achieved in a short time
  9. Keywords:
  10. Sharing Economy ; Service Composition ; Scheduling ; Alternating Direction Method of Multipliers (ADMM)Algorithm ; Blockchain ; Cloud Manufacturing Scheduling ; Decentralized Strategy

 Digital Object List

 Bookmark

No TOC