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

Fog computing has emerged as a promising paradigm to extend cloud services closer to IoT devices, improving response times & reducing network congestion. However, efficient load balancing in fog computing is essential to maximize performance, reduce costs, ensure energy efficiency, & maintain a high quality of service, ultimately supporting the demands of latency-sensitive & resource-intensive IoT applications. The primary objective of task mapping in computing environments such as cloud, fog, & edge computing is to allocate tasks across available resources in an efficient & effective manner, particularly in fog computing, where resources are distributed & closer to end devices. This paper presents a hybrid approach that integrates Genetic Algorithm (GA) & Deep Q-Learning (DQL) for task mapping in fog computing environments. The objective is to minimize makespan & computational costs while maintaining high resource utilization. Our approach leverages a GA to perform initial task allocation by exploring a broad solution space, thereby enhancing convergence toward optimal scheduling patterns. This solution is refined using DQL, which adapts to dynamic environments by learning from continuous feedback and enabling real-time decision-making. By combining the exploration strengths of GA with the adaptive capabilities of DQL, the proposed method effectively manages task allocation & resource utilization. Experimental results show that our hybrid approach outperforms baseline methods, significantly reducing makespan & operational costs.