Hybrid GA-DQL approach for efficient task mapping of IoT applications in fog computing framework.

Area of Science:

• Computer Science
• Artificial Intelligence
• Distributed Systems

Background:

• Fog computing extends cloud capabilities closer to IoT devices, necessitating efficient load balancing for performance and cost-effectiveness.
• Effective task mapping is crucial in fog environments to manage distributed resources and meet demands of latency-sensitive IoT applications.
• Current task mapping strategies may not fully address the dynamic and resource-constrained nature of fog computing.

Purpose of the Study:

• To develop and evaluate a hybrid task mapping approach for fog computing environments.
• To minimize makespan and computational costs while maximizing resource utilization.
• To enhance the efficiency and adaptability of task allocation in distributed fog infrastructures.

Main Methods:

• A hybrid approach integrating Genetic Algorithm (GA) for initial broad exploration and Deep Q-Learning (DQL) for adaptive refinement.
• GA is used for initial task allocation to explore the solution space and converge towards optimal scheduling patterns.
• DQL refines the allocation by learning from dynamic feedback, enabling real-time decision-making and adaptation.

Main Results:

• The proposed hybrid GA-DQL approach significantly outperforms baseline methods in task mapping.
• Demonstrated reduction in makespan and operational costs compared to existing strategies.
• Achieved high resource utilization through effective task allocation and adaptive learning.

Conclusions:

• The hybrid GA-DQL method offers a robust solution for task mapping challenges in fog computing.
• This approach effectively balances exploration and exploitation for optimal resource management in dynamic environments.
• The findings support the adoption of integrated AI techniques for enhancing fog computing performance and efficiency for IoT.