Abstract The exponential growth in user demand, heterogeneous applications, and ultra-low latency requirements of next-generation wireless communication has propelled the need for a paradigm shift in task offloading and resource management strategies, particularly within 6G networks. This paper proposes an intelligent hierarchical Edge-Fog-Cloud computing architecture that efficiently manages computational resources and optimizes task offloading decisions in real-time. Leveraging the strengths of reinforcement learning (RL) for dynamic decision-making and federated intelligence for privacy-preserving collaboration, the framework addresses the key challenges of latency reduction, energy efficiency, and resource utilization under constrained network environments. The proposed architecture distributes computational workloads among edge, fog, and cloud layers based on real-time context awareness, user mobility, and service-level agreements (SLAs). A deep Q-learning (DQL) model dynamically learns optimal offloading policies, while a federated averaging scheme ensures that data remains decentralized, mitigating security and privacy risks. The decision engine incorporates factors such as bandwidth availability, queue length, CPU cycles, and energy constraints to select the most efficient execution layer for each user task. Comprehensive simulations, performed using a customized 6G network emulator, demonstrate that our hybrid edge-fog-cloud system outperforms traditional centralized and two-tier architectures. Results indicate up to 42.3% reduction in average latency, 36.8% improvement in energy efficiency, and 22.5% increase in task success rate. Furthermore, the federated approach reduces communication overhead by 29% compared to centralized learning schemes. This study marks a significant contribution toward the realization of intelligent, decentralized 6G ecosystems by integrating multi-layer computation, autonomous task offloading, and privacy-preserving collaborative learning. The findings pave the way for future research on self-evolving edge infrastructures and ultra-reliable low-latency communications (URLLC) in mission-critical applications such as autonomous vehicles, remote surgeries, and metaverse computing.
P. Sundaravadivel (Thu,) studied this question.