Combining Fog Computing with Sensor Mote Machine Learning for Industrial IoT

Area of Science:

• Computer Science
• Electrical Engineering

Background:

• Digitalization and the Industrial Internet of Things (IIoT) necessitate efficient data communication.
• Transmitting raw sensor data to cloud back-ends consumes significant spectrum and energy.
• Fog computing offers a potential solution for localized data processing.

Purpose of the Study:

• To investigate the benefits of fog computing for Industrial Internet of Things applications.
• To propose a novel distributed learning model on sensor devices.
• To reduce data transmission and conserve energy and spectrum.

Main Methods:

• Developed a distributed learning model on sensor devices.
• Simulated data streams in a fog computing layer instead of transmitting raw data.
• Communicated updated model parameters at longer intervals to the fog system.
• Implemented and tested the framework in a real-world testbed.

Main Results:

• Achieved a 98% decrease in wireless packets sent.
• Maintained 97% accuracy in data stream simulation at the fog node.
• Observed an end-to-end delay of 180 ms in the three-layer framework.

Conclusions:

• Fog computing combined with distributed sensor-level modeling benefits IIoT applications.
• The proposed framework effectively conserves spectrum and energy.
• This approach offers a viable solution for efficient data processing in wireless sensor networks for IIoT.