Cloud systems are inefficient in processing delay-sensitive applications due to the WAN latency associated. To augment the processing of cloud services and provide delay-free computation, fog computing is used. The delay sensitivity of the tasks and heterogeneity of the fog-cloud hybrid architecture calls for efficient resource allocation policies. The decision making must be precise and also multiple criteria must be considered while deciding which resources to allocate. In this article, we propose two variants of analytic hierarchy process (AHP)-based resource allocation policies for fog-cloud hybrid systems. The proposed resource allocation policies consider network load, in addition, to the compute load during decision making. The overall aim of the resource allocation policies is to reduce the delay incurred by each task. The allocation policies differ in the way they assign weights to each criterion of optimization. One of the resource allocation policies uses predetermined weights for compute and network while the second method finds the weights dynamically from the overall data. The experimental results show that the proposed approach outperforms existing resource allocation approaches thereby showing the usefulness of AHP-based optimization in fog-cloud hybrid systems.

中文翻译：

---

使用多准则AHP技术的雾云混合系统中的动态资源分配

由于相关联的WAN延迟，云系统在处理对延迟敏感的应用程序方面效率低下。为了增强云服务的处理并提供无延迟的计算，使用了雾计算。任务的延迟敏感性和雾云混合架构的异构性要求有效的资源分配策略。决策必须精确，并且在决定分配哪些资源时还必须考虑多个标准。在本文中，我们为雾云混合系统提出了两种基于层次分析法（AHP）的资源分配策略。所提出的资源分配策略在决策过程中还考虑了网络负载以及计算负载。资源分配策略的总体目标是减少每个任务引起的延迟。分配策略将权重分配给每个优化标准的方式不同。资源分配策略之一将预定的权重用于计算和网络，而第二种方法则从整个数据中动态找到权重。实验结果表明，所提出的方法优于现有的资源分配方法，从而证明了基于AHP的优化在雾云混合系统中的有用性。