FogFrame: a framework for IoT application execution in the fog.

Area of Science:

• Computer Science
• Distributed Systems
• Cloud Computing

Background:

• Existing fog computing research lacks practical frameworks for real-world implementation.
• Managing edge and cloud resources for Internet of Things (IoT) applications presents significant challenges.

Purpose of the Study:

• To design and implement FogFrame, a concrete framework for managing and monitoring fog computing landscapes.
• To enable decentralized service placement, deployment, and execution for IoT applications within fog environments.

Main Methods:

• Formalized a system model for service placement with an objective function and constraints.
• Implemented greedy and genetic algorithms for decentralized service placement.
• Evaluated FogFrame using a real-world testbed, focusing on Quality of Service and resource utilization.

Main Results:

• FogFrame adapts service placement to demand and available resources.
• Greedy placement maximizes edge utilization; genetic algorithm balances cloud/edge to prevent overload.
• Edge service deployment is 14% faster than cloud deployment.
• Genetic algorithm accommodates new applications better, maintaining ~50% CPU utilization on edge devices.

Conclusions:

• FogFrame effectively manages fog landscapes and executes IoT applications.
• The genetic algorithm offers superior adaptability for new applications and load balancing.
• The framework demonstrates robust reaction to runtime events like device failures and overloads through service migration.