Enhancing smart factory performance via hybrid scheduling and intelligent resource management

Area of Science:

• Computer Science
• Artificial Intelligence
• Industrial Engineering

Background:

• The proliferation of connected devices generates vast data, posing challenges for scalable, private, low-latency solutions.
• Smart factories and Industrial Internet of Things (IIoT) require efficient data processing closer to the source.
• Resource-limited edge devices face latency issues with simultaneous multi-model inferences.

Purpose of the Study:

• To propose a fog computing approach for enhancing machine-to-machine architectures in smart manufacturing.
• To address the challenge of high latency in multi-model inferences on edge devices.
• To develop an AI-driven scheduler for real-time optimization in Industry 4.0.

Main Methods:

• Implemented a fog computing system with container-based orchestration for autonomy and peer-to-peer communication.
• Utilized lightweight deep learning algorithms for simultaneous model inference on edge devices.
• Introduced a hybrid partial swarm optimization-genetic algorithm scheduler to minimize task initiation latency.

Main Results:

• The hybrid scheduler dynamically optimizes task initiation times, reducing overall latency.
• The system integrates IoT and digital twins for adaptive, real-time optimization.
• Achieved significant gains in agility and performance for smart manufacturing applications.

Conclusions:

• The proposed fog computing approach with a hybrid AI scheduler effectively reduces latency in IIoT edge devices.
• This AI-driven model supports adaptive, real-time optimization, balancing multiple objectives for Industry 4.0.
• The innovation lies in enhancing efficiency and performance within complex smart manufacturing environments.