Double Auction-Based Manufacturing Cloud Service Allocation in an Industrial Park
Kai Kang,Su Xiu Xu,Ray Y. Zhong,Bing Qing Tan,George Q. Huang
DOI: https://doi.org/10.1109/tase.2020.3029081
IF: 6.636
2022-01-01
IEEE Transactions on Automation Science and Engineering
Abstract:Industrial parks are regarded as a possible fashion to reduce the manufacturing costs and increase the efficiency by sharing cost-effective infrastructure and communal services. However, the waste of manufacturing resources could impede the development of manufacturing enterprises that locate in an industrial park when market demands fluctuate dynamically. This article aims to use the cloud manufacturing (CMfg) to relieve the temporary shortage of manufacturing resources and capabilities by sharing them among these enterprises in an industrial park. This article focuses on manufacturing cloud service allocation (MCSA), which is one of the critical processes to implement CMfg. Two multiunit double auction mechanisms are proposed to allocate the manufacturing cloud services (MCSs) with reasonable trade price in bilateral markets, and their properties and bidding strategies are proved. Numerical studies are conducted, and the results show the effectiveness and efficiency of two mechanisms for dealing with MCSA and verify the theoretical proofs. Note to Practitioners—MCSA in CMfg is a complicated and challenging task. The current industrial applications rely on optimization methods, such as all-in-one methods (e.g., genetic algorithm) and multidiscipline design optimization methods (e.g., analytical target cascading), to allocate MCSs. While these methods can help cloud operators obtain feasible allocation solutions rapidly, there exist several disadvantages: 1) they can only solve problems in one-sided settings (one customer with many providers); 2) there is an assumption that the customer's demand can be provided by any providers, which means providers' capacities are infinite; and 3) the problem is optimized based on the fixed prices provided by providers, without considering supply and demand. This article presents two multiunit double auction mechanisms that allow mul-iple customers to trade with multiple providers. Allocation rules can break the assumption and providers' capacities are limited. MCSs are also priced dynamically based on supply and demand. We show how cloud operators apply two mechanisms first and demonstrate their properties and bidding strategies of customers and providers when the mechanism is not incentive compatible. Finally, our experiments and analysis fully validate the applicability and efficiency of two mechanisms and illustrate how cloud operators can maximize their utilities.
automation & control systems