Distributed Agent-Based Orchestrator Model for Fog Computing

Area of Science:

• Computer Science
• Distributed Systems
• Network Engineering

Background:

• Fog computing extends cloud services to the network edge, closer to end-users.
• Task placement in fog networks is challenging due to dynamic resources and mobile devices.
• Existing algorithms often lack support for mobility, resource variability, and fault recovery.

Purpose of the Study:

• To propose a distributed, agent-based orchestrator model for flexible service provisioning in dynamic fog environments.
• To address the service or fog application placement problem (SPP, FAPP) considering fog node constraints.
• To enhance resilience, scalability, and real-time performance in fog networks.

Main Methods:

• Developed a distributed agent-based orchestrator model for fog application placement.
• Incorporated constraints such as CPU, memory, battery level, and security.
• Implemented and tested a prototype system using real hardware.

Main Results:

• The proposed model demonstrated efficiency in response latency and computational overhead.
• The distributed decision-making approach enhanced scalability and resilience.
• The system effectively managed dynamic fog environments with mobile users.

Conclusions:

• The distributed orchestrator model is suitable for fog network applications requiring scalability, mobility, and fault tolerance.
• Decentralized decision-making improves load distribution and system resilience.
• The approach offers an efficient solution for the dynamic service placement problem in fog computing.