Article Preview
TopIntroduction
Recent advancement in wireless technologies finds the opportunities for several Internet of Things (IoT) based resources like wearable devices, home appliances, and software to communicate information with each other (Li et al., 2015). Since the domain of applications is vast in IoT as compared to conventional techniques that impose the challenge of achieving better Quality of Experience (QoE). It is the user's subjective satisfaction with the application which becomes more complicated for network operators to perceive. Some of the quality issues such as Quality of Service (QoS) and QoE have a huge impact on the successful adoption of IoT based applications. Though QoE is a subset of QoS, there are certain dissimilarities between them which point towards different policies based on service management (Varela et al., 2014). Services like cloud computing storage or computer network fall under the QoS category which is defined as the amount of overall performance of the service.
On the other hand Fog computing acknowledges the situation of latency with cloud services and proposes a new paradigm. Fog computing also shifts the storage and computing to the Fog nodes located at the network edge which in turn improves the overall efficiency, latency, and QoE (Brogi & Forti, 2017). Here, QoE focuses on the end-users perspective of the quality of an application or a service that minimizes latency in the network. Thus, a platform extending cloud-based utilities at the edge of the network for IoT based applications is required and this is termed as Fog computing. This platform improves QoE for a wide range of IoT-based applications like mobile, latency-sensitive, and geo-distributed applications (Mahmud, Kotagiri, et al., 2018). Fog computing includes several layers in its architecture and each layer has different computation power and latency which are required by a different set of applications to achieve high QoE and QoS. Fog computing has emerged as a supporting model for the storage and computation requirements of IoT based applications. Along with cloud data centers, Fog offers various services to the end-user in a real-time environment but the QoE requirement may vary from application to application. A relationship between latency and computation power that varies from cloud computing to Fog computing is shown in Figure 1.
Figure 1. Latency and computation in multilayer Fog
Figure 1 includes the basic parameters like storage, maximum latency, CPU cycle, network bandwidth, and processing time that vary significantly in different layers of Fog, and because of this application placement in Fog is one of the challenging tasks. Therefore, effective application placement policies are required to achieve the benefits of the Fog computing environment. Fog computing services are presented at the edge of the network that also decreases the latency of the services, enhances QoS which contributes towards a greater experience for end-users. Fog nodes can virtualize their applications and thus it can run applications in the multitenancy model. For example, QoE that is received by the user while playing video games can be improved by placing a specific task at the desired level in Fog computing architecture (Dastjerdi et al., 2016). Each application needs to be placed on an appropriate Fog level which is according to the application parameters so that the QoE at the end-user level could be achieved. In this paper, authors have used the Analytical Hierarchical Process (AHP) technique in a distributed computing environment and proposed a framework for Multi-Criteria Decision Making (MCDM) (Mu & Pereyra-Rojas, 2018). AHP uses paired comparisons ratio scales using which input values can be derived from subjective opinions such as preference and real measurement like weight, price, etc. Initially, AHP forms a class of the problem, and then it provides the cost at each level of the class and forms a pairwise comparison matrix. Further, this paper focuses on the development of a framework that helps in allocating the application to the appropriate Fog layer with the desired QoE.