Towards Decentralized Fog Computing: A Comprehensive Review of Models, Architectures, and Services

Cloud computing, in general terms, is a multi-tenant architecture that serves as a virtual platform and storage solution. Such solutions offer various services such as static audio/video storage, video streaming on-demand, that requires constant push–pull of resources to-and-fro the client network and cloud, leading to bandwidth bottleneck and network latency. In an evolution to optimize the network usage metrics, a novel decentralized architecture ‘Fog Computing’ has emerged. It is an extension to cloud and enables the deployment of computing resources at the network edge, providing low-latency, high-bandwidth services to end-users. This paper presents arguments to deploy fog network for resource constraint systems using Software-defined models and styles—hierarchical, flat, and understand the network usage improvements per style. For resource-demanding systems, the paper proposes to opt-in for HWP and FNP style, and HNP, FWP on the other hand for latency-agnostic ones. It also covers architectural dimension analysis and a decision matrix for the researchers, to implement an operational model for various applications, smart cities, health care, vehicular networks, IOT, etc. An additional distinguishing artefact that this paper unfolds is the literary contributions around device dimension, which is a key dimension in designing Fog networks. In summary, the paper suggests leveraging the benefits of fog paradigm and understanding the architectural patterns to set up a fog network for application-specific needs.