Generalized Dynamic Multilayer Fog Computing Architecture

Abstract

The concept of Smart city and Smart Home is a promising and challenging one which defines the future of urban infrastructure development by integrating the Internet of Things (IoT) with Cloud Computing. Fog Computing with its latest framework on multilayering where each layer focuses on the different aspect of the architecture has proven useful in its goal to reduce space, time and computation overhead from cloud perspectives, due to its traffic offloading and edge device proximity for computation. But, a need for platform generalization and dynamic behavior still exists as most of the smart home needs to operate in multiple environments whereas the existing system mainly works on a rigid model with single infrastructure support. In this paper, a modified version of multilayer fog architecture has been proposed with scope for generalization and dynamic operation in terms of data management and placement. The proposed system consists of five sections - the Lower Fog Layer for end device collection operation and prioritization, Dockerized Middle Layer for node-specific operations like event detection and filtered data propagation, ICFN (Interconnecting Fog Node) Layer for interconnecting fog nodes, Cloud Server for data analytics and Fog Maintenance Remote Server for platform customization. The additional platform customization and multiple service support in the proposed architecture have also made significant improvements with regards to data distribution, request bandwidth, and request failure rate.