| Abstrakt: |
Today, Fog computing plays an important role in minimizing service delivery delays and network relaxation. So many architectures with different purposes are presented in the Fog environment. One of the major drawbacks of most of these architectures is the hierarchical process of servicing requests. Applications are sent to the Fog environment and if Fog is unable to provide the service, it is sent to the Cloud. This process of service is somewhat time-consuming and not suitable for real-time requests and with increasing the number of requests and the number of Fog nodes, choosing the appropriate node to place the application becomes a challenge. In this article to solve these challenges, a framework is provided and some modules are considered that the application specifications are first sent to these modules. These modules contain information about the Fog and Cloud environments, which based on the application specifications in the shortest possible time, determine whether the request is sent to the Cloud or Fog and then the application is sent for service. Therefore, the application that must be served in the Cloud is not entered into the Fog environment. To properly manage the application placement in Fog, some domains are provided for fog nodes, and using a module the most appropriate node for servicing the request is determined. In this framework, service response times and SLA violations are greatly reduced and increasing requests or increasing Fog nodes does not pose a problem in service delivery. The proposed method is implemented in the iFogSim simulation environment with FCFS and PSO algorithms and its results have been evaluated with valid articles. It was observed that the proposed method, on average, reduces the response time by 74.16% compared to the state-of-the-art research works, which is a significant improvement in service response time. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink