An Intelligent CNN-Based QoS-Aware Resource Scheduling Approach for Fog Computing

Fog computing has emerged as a promisingparadigm for supporting latency-sensitive and data-intensiveInternet of Things (IoT) applications by extending cloud servicescloser to the network edge. However, the heterogeneous anddynamic nature of fog environments makes efficient resourcescheduling a challenging task, particularly in meeting Quality ofService (QoS) requirements such as low latency, high reliability,and energy efficiency. To address these challenges, this paperproposes an intelligent CNN-based QoS-aware resource scheduling approach for fog computing. The proposed method leveragesthe automatic feature extraction and predictive capabilities ofConvolutional Neural Networks (CNNs) to make adaptive taskto-resource allocation decisions under dynamic workloads. Thescheduling strategy aims to minimize task latency and systemenergy consumption while ensuring QoS constraints are satisfied.Extensive simulations conducted using iFogSim with varying taskloads and fog node configurations demonstrate that the proposedapproach significantly outperforms conventional meta-heuristicalgorithms, including Genetic Algorithm (GA), Particle SwarmOptimization (PSO), and Artificial Bee Colony (ABC). Experimental results show that the CNN-based scheduler achieves up to80% reduction in energy consumption, lower execution latency,and a consistent 100% QoS satisfaction rate under high workloadconditions. These results validate the effectiveness of the proposedapproach as a scalable and energy-efficient solution for nextgeneration fog-enabled IoT systems.