The fast development in IoT and Cloud technologies has propelled the emergence of a variety of computing paradigms, among which Fog and Edge computing are salient computing technologies. Such new paradigms are opening up new opportunities to implement novel application scenarios, not possible before, by supporting features of mobility, edge intelligence and end-user support. This, however, comes with new computing challenges. One such challenge is the allocation of applications to Fog and Edge nodes. Indeed, for some application scenarios larger computing capacity might be needed. Therefore, due to co-existence of computing devices of different computing granularity, techniques for grouping up and clustering resources into virtual nodes of larger computing capacity are required. In this paper we present some clustering techniques for creating virtual computing nodes from Fog/Edge nodes by combining semantic description of resources with semantic clustering techniques. Then, we use such clusters for optimal allocation (via heuristics and Liner Programming) of applications to virtual computing nodes. Simulation results are reported to support the feasibility of the model and efficacy of the proposed approach. First Fit Heuristic Algorithm (FFHA) outperformed ILP method for medium and large size instances. Likewise, FFHA performed more consistently than ILP on various experimental setting. Finally, the results showed that the proposed clustering techniques deliver relatively fast response times, while enabling the service of a larger number of applications, with more demanding requirements.

中文翻译：

---

通过语义聚类技术将应用程序分配给雾资源：使用智能交通系统的场景

物联网和云技术的快速发展推动了多种计算范式的出现，其中雾计算和边缘计算是突出的计算技术。通过支持移动性、边缘智能和最终用户支持等特性，这种新范式为实现新的应用场景开辟了新的机会，这在以前是不可能的。然而，这带来了新的计算挑战。其中一项挑战是将应用程序分配给 Fog 和 Edge 节点。事实上，对于某些应用场景，可能需要更大的计算能力。因此，由于不同计算粒度的计算设备的共存，需要将资源分组和集群为更大计算能力的虚拟节点的技术。在本文中，我们通过将资源的语义描述与语义聚类技术相结合，提出了一些从雾/边缘节点创建虚拟计算节点的聚类技术。然后，我们使用这样的集群将应用程序优化分配（通过启发式和线性编程）到虚拟计算节点。报告模拟结果以支持模型的可行性和所提出方法的有效性。First Fit Heuristic Algorithm (FFHA) 在中型和大型实例上的表现优于 ILP 方法。同样，FFHA 在各种实验环境中的表现比 ILP 更一致。最后，结果表明，所提出的集群技术提供了相对较快的响应时间，同时能够为具有更高要求的大量应用程序提供服务。