Edge Computing: A Review on Computation Offloading and Light Weight Virtualization for IoT Framework

Edge Computing: A Review on Computation Offloading and Light Weight Virtualization for IoT Framework

Minal Parimalbhai Patel (Computer Engineering Department, A. D. Patel Institute of Technology, Gujarat Technological University, Gujarat, India) and Sanjay Chaudhary (School of Engineering and Applied Science, Ahmedabad University, Gujarat, India)
Copyright: © 2020 |Pages: 11
DOI: 10.4018/IJFC.2020010104
OnDemand PDF Download:
No Current Special Offers


In this article, the researchers have provided a discussion on computation offloading and the importance of docker-based containers, known as light weight virtualization, to improve the performance of edge computing systems. At the end, they have also proposed techniques and a case study for computation offloading and light weight virtualization.
Article Preview

1. Introduction

In the current trend of technology (Wang & Alexander, 2016), the emergence of IoT is considered for enabling real-world applications and it has been justified by following technologies such as sensors and embedded systems, ultra-low power based processors, Radio Frequency Identification (RFID), mobile services, cloud and fog computing, wireless communication etc. The computation off-loading is performed using cloud and fog technologies for managing large-scale data analysis and managing huge operations.

Figure 1.

Fundamentals of IoT Edge Computing (Premsankar, Di Francesco, & Taleb, 2018)


The questions raised by IoT designer is so challenging and there may be multiple solutions exist: i) trade-off between quality of service and energy consumption, ii) off-load data for computation and storage or consider on-board processing, iii) which communication technology is applied under certain requirements to bring IoT system more adequate for real-world operations? iv) requirement analysis for relevant range of IoT devices for communication, considering data-rate and low-power devices constraints etc. The data generated from IoT devices including audio, video or unstructured data is processed using big-data approach. In Figure 1, the IoT edge computing layers are shown to manage the services from cloud to smart devices.

The different computing mechanism (Premsankar, Di Francesco, & Taleb, 2018) for IoT edge computing is discussed below:

  • Device level: This mechanism is used mainly for low-power requirements and the major decision is required to perform computation on device itself or to offload it for better computation;

  • Gateway level: It is also known as smart phone centric approach. It is used for those devices which require more computational power and useful for healthcare and engineering applications. It is able to manage the data communication through the wireless communication and the issues concern with latency is required to minimize for better performance;

  • Fog level: This layer is able to give more computation power compared with device and gateway approaches. It is a micro cloud activity to manage the data closer to the user and it is able to solve data analysis at greater depth;

  • Compare to cloud computing: It is used to reduce latency and bandwidth issues for different IoT applications;

  • Cloud level: It is mainly used for server processing at cloud and big data can be processed to provide decision making in different cloud layers.

In Figure 2, three different edge computing platforms (Premsankar, Di Francesco, & Taleb, 2018) are shown and the comparison of edge system to the fog and cloud systems are shown with necessary elements.

Figure 2.

Edge Computing platforms (Premsankar, Di Francesco, & Taleb, 2018)


2. Iot-Enable Middle Layer Technologies

In this section, three different IoT-enable middle layer technologies are discussed below:

  • 1.

    Fog Computing

Complete Article List

Search this Journal:
Open Access Articles: Forthcoming
Volume 4: 2 Issues (2021): Forthcoming, Available for Pre-Order
Volume 3: 2 Issues (2020)
Volume 2: 2 Issues (2019)
Volume 1: 2 Issues (2018)
View Complete Journal Contents Listing