The architecture of fog network - A bridge between Cloud and IoT (Part 2)
According to the previous part of the blog, we have briefly touched the concept of Fog computing. We discussed about the characteristics of Cloud based IoT networks that led to the creation of fog layer in part 1. In this part, we will look into the internal layers of a Fog network that are essential for its functioning.
Fog is a virtualized platform in an IoT network that provides services of computation, storage and networking, resources to the end Things from the Cloud. According to Open Fog Consortium [1], the use cases of Fog computing for the IoT systems are,
- Security
- Cognition : Client-centric objectives
- Agility
- Latency
- Efficiency
The services of Fog is distributed between the core network (of routers that use IP/MPLS core), access nodes (FANs that use 3G/4G/Ethernet) and the gateways of a network
Fog computing consists of high scalability where different kinds of applications with various data flow can be combined and processed. This scalability provides the following characteristics of Fog network.
- virtualization - Providing a virtual platform for computing
- orchestration - The coordination and management of Fog nodes and services
- multi tenancy - Secured cloud usage by multiple clients
- modularity - separable interconnected components of the network
- programmability - Usage of APIs for operating in various platforms & environments, supporting multiple versatile applications
Muhammad Aazam et al [2] defined a layered architecture for Fog computing based on its functionalities which are transporting data, security, temporary storage, preprocessing, monitoring and virtualization.
Software based infrastructure of Fog
In order to cope with the requirements of different applications, Fog Computing was proposed with a virtualized infrastructure which contains a policy-driven orchestration layer for resource management of Fog nodes.
The architecture of Fog Computing uses APIs to exhibit its services to applications for their development and deployment. The infrastructure of Fog provides these APIs in its Abstraction layer and Orchestration layer for scalability among the nodes of the network [3].
- Physical Resources
Physical resources of fog infrastructure describe the heterogeneous devices that are connected with fog nodes for various application purposes. The devices can be in wide range from unintelligent to intelligent such as sensors, gateways, APs, routers, mobile phones, servers etc. Due to this diversity, the resources needed by these end devices also vary majorly.
The devices differ in the key resources like storage capabilities, hardware structure, linking speed, network accessing technology, operating systems and application software according to their functionalities.
- Abstraction Layer
The Abstraction layer of Fog Computing is a virtualized layer that is dedicated for the main services of Fog that are storage, computation and network. The layer provides Programmable Interfaces with consistency for monitoring and controlling of physical resources, services despite of their diversity. The techniques used for virtualization in this layer enable multiple users in the network to utilize its services.
- Orchestration Layer
Orchestration Layer of Fog Computing is the responsible layer for managing the dynamic, policy driven life cycle of fog nodes. The layer contains several components to contribute its functionalities.
- Distributed storage for Meta data, rules and policy storing.
- Policy engine for decision making on enabling services.
- Software agents called ‘Foglets’ in each node to monitor the nodes through Abstraction APIs.
- Messaging bus for exchanging control messages to implement orchestration in Fog nodes.
The orchestration layer not only does the network and resource management, but also enhances the scalability of the fog network.
According to the above mentioned architecture and infrastructure of Fog, the main features of Fog computing distributes its services (network, storage, computation) to end devices. Even though, Fog paradigm is an extension of Cloud of Things which can be accompanied or replaced by Hybrid cloud, Intelligent Edge and CDNs (Content Delivery Networks) to overcome the barriers of Cloud [4].
Up to now, we learnt the architecture and functioning of the layers in the Fog network. The next part of the blog briefly describes the security issues of Fog computing and some of the proposed techniques to overcome those security issues.
References
[1]. “OpenFog Consortium |”, Openfogconsortium.org, 2018.[Online].Available:https://www.openfogconsortium.org/. [Accessed: 18- Apr- 2018].
[2]. M. Aazam and E.-N. Huh, “Fog Computing: The Cloud-IoT\/IoE Middleware Paradigm,” IEEE Potentials, vol. 35, no. 3, pp. 40–44, 2016.
[3]. Bonomi, Flavio, et al. “Fog Computing: A Platform for Internet of Things and Analytics.” Big Data and Internet of Things: A Roadmap for Smart Environments Studies in Computational Intelligence, 2014, pp. 169–186., doi:10.1007/978–3–319–05029–4_7.
[4]. C. Byers, “Fog computing Architecture and why the Cloud/smart devices are not enough- Perspective Cisco Industry”, San Jose, United States, 2014.
