A software defined-based hybrid cloud for the design of smart micro-manufacturing system

Abstract

For purposes of re-configurability and flexibility, a few control data are extracted from manufacturing data generated from a production line in a smart factory. A production manager can rearrange those control data selected from the manufacturing data to formulate the mission of a production line. All related data are saved in a cloud storage after being verified by a security mechanism. They can be accessed only with permission. Since the structures of control data and manufacturing data are different, they are saved in various databases. However, accessing different databases results in additional communication cost, and the data-save performance will be decreased simultaneously. The production line setting may be changed based on the mission of the received orders, so the control data will be modified. It costs huge communication overhead if the cloud storage queries the control data for each request. In this paper, we propose two cache-based mechanisms, termed laziness approach and flow-based update (FBU) approach, to reduce the cost of verifying the save permission. The laziness approach gets the corresponding control data when the received data can not be matched to the information in the cache. The update process of the FBU approach is similar to that of the laziness, but the FBU downloads the entire control data of the specific production line. According to our analysis results, both mechanisms provide better performance than that of on-demand approach in terms of data-save process. In the worst case analysis, the FBU approach only needs a half cost of that required by the laziness approach. Moreover, the optimal cache size is inversely proportional to the stability of the production line setting, and we also suggest an optimal setting of the cache size.