A Scalable Two-Hop Multi-Sink Wireless Sensor Network for Data Collection in Large-Scale Smart Manufacturing Facilities

Abstract

In industrial fields, wireless sensor networks have been massively deployed for the pur-pose of data collection. For the various application scenarios of smart manufacturing in Industry 4.0, versatile production tasks demand dynamic features both in production lines and manufacturing processes. Therefore, the design and performance of the corresponding data collection mechanisms are facing unprecedented challenges. In this work, we propose a unified data description and management framework. This framework possesses high flexibility that it is able to identify an unknown data type and accord an adequate description. Besides, the scalability of this framework enables the provision of handy interfaces for the exploitation of stored data. Then, we develop two network connectivity models in one dimension and two dimensions. These two models greatly facilitate the measurement of the level of connectivity for a wireless sensor network. At last, we elaborate a two-hop multi-sink routing scheme to alleviate the flooding problem. This scheme contains a novel r-Kruskal algorithm for the sink nodes and an efficient two-hop routing method for the whole network. The flooding effect can be neatly controlled with the two-hop scheme. Extensive experiments are conducted to evaluate our proposal. Simulation results show that our model has excellent adaptability to the scale of the network and possesses satisfactory performance in terms of both message overhead and data availability.