The proliferation of ubiquitous Internet of Things (IoT) sensors and smart devices in several domains embracing healthcare, Industry 4.0, transportation and agriculture are giving rise to a prodigious amount of data requiring ever-increasing computations and services from cloud to the edge of the network. Fog/Edge computing is a promising and distributed computing paradigm that has drawn extensive attention from both industry and academia. The infrastructural efficiency of these computing paradigms necessitates adaptive resource management mechanisms for offloading decisions and efficient scheduling. Resource Management (RM) is a non-trivial issue whose complexity is the result of heterogeneous resources, incoming transactional workload, edge node discovery, and Quality of Service (QoS) parameters at the same time, which makes the efficacy of resources even more challenging. Hence, the researchers have adopted Artificial Intelligence (AI)-based techniques to resolve the above-mentioned issues. This paper offers a comprehensive review of resource management issues and challenges in Fog/Edge paradigm by categorizing them into provisioning of computing resources, task offloading, resource scheduling, service placement, and load balancing. In addition, existing AI and non-AI based state-of-the-art solutions have been discussed, along with their QoS metrics, datasets analysed, limitations and challenges. The survey provides mathematical formulation corresponding to each categorized resource management issue. Our work sheds light on promising research directions on cutting-edge technologies such as Serverless computing, 5G, Industrial IoT (IIoT), blockchain, digital twins, quantum computing, and Software-Defined Networking (SDN), which can be integrated with the existing frameworks of fog/edge-of-things paradigms to improve business intelligence and analytics amongst IoT-based applications.

中文翻译：

---

物联网应用的人工智能雾/边缘资源管理：全面回顾、研究挑战和未来展望


医疗保健、工业 4.0、交通和农业等多个领域中无处不在的物联网 (IoT) 传感器和智能设备的激增，产生了大量数据，需要不断增加从云到网络边缘的计算和服务。雾/边缘计算是一种有前途的分布式计算范式，引起了工业界和学术界的广泛关注。这些计算范例的基础设施效率需要自适应资源管理机制来卸载决策和高效调度。资源管理（RM）是一个不平凡的问题，其复杂性是异构资源、传入事务工作负载、边缘节点发现和服务质量（QoS）参数同时产生的结果，这使得资源的效率更具挑战性。因此，研究人员采用基于人工智能（AI）的技术来解决上述问题。本文通过将雾/边缘范式中的资源管理问题和挑战分类为计算资源配置、任务卸载、资源调度、服务放置和负载平衡，全面回顾了雾/边缘范式中的资源管理问题和挑战。此外，还讨论了现有的基于人工智能和非人工智能的最先进解决方案，及其 QoS 指标、分析的数据集、局限性和挑战。该调查提供了与每个分类的资源管理问题相对应的数学公式。 我们的工作揭示了无服务器计算、5G、工业物联网 (IIoT)、区块链、数字孪生、量子计算和软件定义网络 (SDN) 等尖端技术的有前途的研究方向，这些技术可以与现有技术集成雾/边缘事物范式框架，用于改进基于物联网的应用程序中的商业智能和分析。