Resource and Service Management for Fog Infrastructure as a Service

Abstract

Cloud computing stops short in its offerings towards deployment of latency-critical and bandwidth-intensive applications. Fog computing is emerging as complementary to cloud computing in realizing the deployment of large-scale Internet of Things (IoT) environments for such applications. It supports geographically dispersed IoT devices and users leveraging compute nodes of varied resource capacities in vicinity of the devices/users. Researchers have demonstrated the necessity of fog computing towards deployment of latency-critical and bandwidth-intensive applications. Fog computing infrastructure is recommended for hosting the new generation application environments such as autonomous vehicles, emergency services and personalized healthcare. The Fog Infrastructure as a Service (FIaaS) model facilitates leasing of shared infrastructure resources such as compute, network, and storage on fog nodes which are deployed by Fog Service Providers. Several research problems must to be solved prior to the real-world deployment of large-scale fog computing environments. This research identified and attempted to solve some of them in the FIaaS model. It has developed an event-driven simulator, PFogSim, to facilitate the simulation of large-scale, dynamic, and mobile fog computing environments comprising thousands of devices with defined multi-layered hierarchical structure representing smart city deployments using IoTs. Towards efficient management of fog, this research proposes a Hierarchical and Autonomous Fog Architecture (HAFA) to organize heterogeneous fog nodes into multi-layered connected hierarchy based on several parameters such as location, distance from IoT devices and/or users, resource configuration, privacy and security. It groups fog nodes to facilitate resource pooling and local control, and logically links such groups to facilitate disaster readiness and autonomy. Leveraging HAFA, this research proposes and tests a distributed approach to select a cost-efficient fog node to host any given application service among prospective fog nodes with available infrastructure resources considering both computation and communication costs. To cater to the unique needs of the fog environment, it also develops a novel location and network-aware approach for service pricing in FIaaS environments.