Scholarlink

<p>The increasing scale and heterogeneity of Internet of Things (IoT) deployments have intensified the limitations of centralized cloud computing (CC), particularly in applications requiring low latency, efficient bandwidth utilization, and localized decision-making. Fog Computing (FC) has emerged as a distributed computing paradigm that places computational and storage capabilities closer to data-generating sources, thereby bridging the gap between edge devices and remote cloud infrastructures. This survey presents a systematic and critical review of FC research spanning the period from 2015 to 2025. Using a Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA) inspired review methodology, the paper synthesizes existing studies to construct a structured taxonomy encompassing architectural models, service abstractions, task scheduling strategies, resource allocation mechanisms, Artificial Intelligence (AI)-enabled optimisation, security solutions, and energy-aware designs. Empirical results reported in the literature indicate that fog-based frameworks demonstrate reported improvements in prior studies cloud-centric systems, achieving latency reductions of up to 70%, bandwidth savings approaching 80%, and energy-efficiency improvements ranging from 15% to 50% across diverse domains including smart healthcare, vehicular networks, smart cities, and industrial IoT. Particular attention is given to learning-driven approaches, including Reinforcement Learning (RL) and Federated Learning (FL), which enable adaptive, scalable, and privacy-preserving fog operations. Based on identified research gaps, we propose FogAI-Optimizer, an integrated conceptual framework synthesized from systematic literature evidence that integrates multiple FC architectural and optimisation dimensions, green AI optimisation, edge intelligence, cloud integration, and dataset-driven validation into a cohesive model for next-generation fog systems. The survey concludes by identifying persistent challenges related to scalability, sustainability, and security, and outlines concrete future research directions, including FogAI-Optimizer implementation, Artificial Intelligence of Things (AIoT) integration, and 6 G-enabled fog deployments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Comprehensive Survey on Fog Computing: Architectures, Techniques, Challenges, and Future Directions

  • 期刊论文

  • 发表日期：

    2026年4月11日

查看全文
  • Prachi Chaturvedi,
  • Shahnawaz Ahmad,
  • Arvind Mewada

摘要

The increasing scale and heterogeneity of Internet of Things (IoT) deployments have intensified the limitations of centralized cloud computing (CC), particularly in applications requiring low latency, efficient bandwidth utilization, and localized decision-making. Fog Computing (FC) has emerged as a distributed computing paradigm that places computational and storage capabilities closer to data-generating sources, thereby bridging the gap between edge devices and remote cloud infrastructures. This survey presents a systematic and critical review of FC research spanning the period from 2015 to 2025. Using a Preferred Reporting Items for Systematic Reviews and Meta Analyses (PRISMA) inspired review methodology, the paper synthesizes existing studies to construct a structured taxonomy encompassing architectural models, service abstractions, task scheduling strategies, resource allocation mechanisms, Artificial Intelligence (AI)-enabled optimisation, security solutions, and energy-aware designs. Empirical results reported in the literature indicate that fog-based frameworks demonstrate reported improvements in prior studies cloud-centric systems, achieving latency reductions of up to 70%, bandwidth savings approaching 80%, and energy-efficiency improvements ranging from 15% to 50% across diverse domains including smart healthcare, vehicular networks, smart cities, and industrial IoT. Particular attention is given to learning-driven approaches, including Reinforcement Learning (RL) and Federated Learning (FL), which enable adaptive, scalable, and privacy-preserving fog operations. Based on identified research gaps, we propose FogAI-Optimizer, an integrated conceptual framework synthesized from systematic literature evidence that integrates multiple FC architectural and optimisation dimensions, green AI optimisation, edge intelligence, cloud integration, and dataset-driven validation into a cohesive model for next-generation fog systems. The survey concludes by identifying persistent challenges related to scalability, sustainability, and security, and outlines concrete future research directions, including FogAI-Optimizer implementation, Artificial Intelligence of Things (AIoT) integration, and 6 G-enabled fog deployments.