IEEE Access (Jan 2020)

Learning-Based IoT Data Aggregation for Disaster Scenarios

  • Min Peng,
  • Sahil Garg,
  • Xiaoding Wang,
  • Abbas Bradai,
  • Hui Lin,
  • M. Shamim Hossain

DOI
https://doi.org/10.1109/ACCESS.2020.3008289
Journal volume & issue
Vol. 8
pp. 128490 – 128497

Abstract

Read online

Industrial Internet of Everything (IIoE), as the deep integration of industry 6.0, the Internet of Things (IoT) and 6G mobile communication technology, pave the way for intelligent industry, enabling industrial optimization and automation. To ensure the high quality of services (QoS) in IIoE, tremendous real-time information generated by the pervasive smart things needs to be aggregated and processed quickly and reliably. However, a large-scale disaster could damage the entire communication network and cut off data aggregation such that Qos is compromised. In this paper, an Intelligent NIB based Data Aggregation Strategy, named (IDAS), is proposed for after disaster scenarios in IIoE. Specifically, IDAS first applies both iterative cubature kalman filter and radial basis function neural network to predict the data collection rates of survived infrastructures. Then, an energy efficient task distribution mechanism is design. Next, a deep reinforcement learning method is developed for the car-carrying NIB route design to perform corresponding task. Eventually, all data are aggregated toward the rescue headquarter by NIB deployment based on Fermat tree constructions. The theoretical analysis and simulations indicate that IDAS is not only energy efficient for after disaster scenarios but requires the least NIB consumption while compared with contemporary strategies.

Keywords