Toward integrated Cloud–Fog networks for efficient IoT provisioning: Key challenges and solutions
Introduction
The era of Internet Technology (IT) has unconsciously migrated to Data Technology (DT) with the proliferation of mobile devices such as smart phones, tablets, wearable devices, etc., the total number of which is predicted to approach 50 billion by 2020 [1]. Billions of these “things” are generating more than two Exabytes of everyday Internet-of-Things (IoT) data featured by 3Vs, i.e. volume, variety, and velocity [[2], [3], [4], [5]]. Provisioning the 3Vs IoT data in the Cloud datacenter networks (DCNs) could be subject to serious technical difficulties, mostly due to the fact that moving all generated IoT data at the devices to the Cloud for analysis would saturate the network capacity and cause unbearable latency, eventually failing the premises of the IoT services.
Specifically, Cloud datacenters (DCs) have been widely deployed by leading IT companies such as Cisco, Google, Microsoft, Amazon, etc., to provide the computing/storage resources to enterprises/individuals under the leasing models of infrastructure-as-a-service (IaaS), platform-as-a-service (PaaS), software-as-a-service (SaaS), etc., [[6], [7]]. Even though Cloud DCs under these leasing models work well for heavyweight and latency-tolerant service requests, they are inherently inappropriate to serve the pervasive 3Vs IoT data with stringent real-time requirements [[8], [9]], mainly due to the long distance between DCs and end-users, in addition to the centralized control and management on DCs. Nonetheless, it was forecasted that by year 2020, up to 92% of workloads should still be processed by Cloud DCs [7]. Cloud DCs will continue to be a significant infrastructure in provisioning the legacy Cloud and the emerging IoT applications.
It becomes apparent that an additional control layer in between the Cloud and the IoT devices situated in the vicinity of the IoT devices for governing the data locality and mobility will be necessary toward the success of provisioning the 3Vs IoT data. With this regard, we have seen extensive research interest shifted from the network core to the edge in the past a few years, generally termed Edge Computing (EC). Based on the concept, many vendor-specific control platforms emerged, including mobile Cloud Computing (MCC), mobile Edge Computing (MEC), cloudlet, etc. [[10], [11], [12]]. As a similar alternative, Fog Computing (FC) has been coined for IoT services by Cisco in 2012 and since then, it has been receiving tremendous attention from academia and industry [[1], [13]].
The key concept behind FC is to take advantage of the end-user devices located at (or near) the edge, which would be rich in IoT resources, i.e. storage, compute, and bandwidth, to process the real-time data of neighboring Things in a one-hop manner so as to minimize the latency. The Cloud and Fog (which is a “Cloud closer to the ground”) are not binary options, in contrast, they are an interdependent and mutually beneficial continuum [14]. For instance, Cloud still coordinates the Fogs and the devices in a Fog as well as handles heavyweight data, nonetheless, the delay-sensitive data can be processed and responded by Fog nodes that are in the vicinity of the IoT devices [[15], [16]].
Even though significant industrial and/or academic progresses have been achieved on either the Cloud or the Edge, respectively, very little progress has been witnessed on an integrated Cloud–Fog framework. Researchers have become aware of this missing piece and have realized the significance of such an integrated architecture, which can optimize the power of both Cloud and Fog to speed up the development of IoT. Research in this field has just recently kicked off, with very little progress being made [17].
In this article, we first provide a literature review of the studies related to the integration of Cloud–Fog networks. We then present iCloudFog, a scalable and agile integrated Cloud–Fog architecture that provisions IoT resources of Cloud and/or Fog to IoT data requests dynamically based on the Cloud/Fog nodes’ availability and capability as well as the IoT data requirements. In the process of constructing the iCloudFog framework, we identify the key challenges being network dimensioning, IoT job scheduling with consideration on QoS, security/privacy, and indoor localization/positioning, and suggest viable approaches to address these challenges.
The remainder of the article is organized as follows. In Section 2, we discuss the differences between Cloud Computing (CC), Edge Computing (EC), and Fog Computing (FC), and then present a literature review. In Section 3, we present the proposed iCloudFog architecture. In Section 4, we discuss they key challenges in iCloudFog such as network dimensioning, resource management/job scheduling, security, and indoor localization/positioning, and present viable solutions. In Section 5, we summarize the article and present the conclusions.
Section snippets
Literature review
In this section, we first briefly discuss the differences between CC, EC, and FC, and then present an overview of the related works. Finally, we discuss the key challenges and issues need to be addressed.
Overview of iCloudFog framework
The proposed iCloudFog architecture is illustrated in Fig. 1. The physical network architecture is composed of the optical backbone interconnect, an access network, and the IoT end-user devices. The abstraction of the physical architecture comprises three layers, namely Cloud, Fog and IoT. With iCloudFog, three types of communication are defined: Cloud-to-Fog (C2F), Fog-to-Fog (F2F), and Fog-to-Thing (F2T).
To ensure high efficiency of the iCloudFog operations, particularly in dealing with data
Key challenges and promising solutions in iCloudFog
The integrated Cloud–Fog (iCloudFog) architecture requires addressing several research challenges so as to enable effective operation. In this section, we focus on four primary ones, namely network dimensioning and configuration, QoS and privacy-aware resource management and job scheduling, security, and positioning/localization. We discuss potential approaches to effectively address these challenges.
Conclusions
Fog computing has been taken as a promising technology for enabling low-latency, flexible and scalable future edge network operation to support intensive IoT data. In this article, we discussed the integration of Fog with the Cloud, which has been shown to have several merits. We then proposed iCloudFog, an integrated Cloud–Fog architecture that enables the construction of different Fog types to fit the different characteristics of IoT devices and data, and Fog nodes. Key research challenges
Acknowledgment
This work is supported by the National Research Foundation of Korea (NRF)grant funded by the Korean government (Grand No. 2015R1C1A1A02036536 and Grand No. 2018R1D1A1B07051118).
Limei Peng received the M.S. and Ph.D. degrees from the Chonbuk National University,South Korea, in 2006 and 2010, respectively. In 2011, she was a Research Professor with Grid Middleware Research Center, Korea Advanced Institute of Science and Technology, South Korea. She has been an Associate Professor with the School of Electronic and Information Engineering, Soochow University, China, for more than two years. She has been an Assistant Professor with the Department of Industrial Engineering,
References (30)
SOVCAN: safety-oriented vehicular controller area network
IEEE Commun.
(2017)- Cisco fog computing solutions: Unleash the power of the Internet of Things, white paper, Cisco,...
TempoRec: temporal-topic based recommender for social network services
Mobile Netw. Appl.
(2017)Deep features learning for medical image analysis with convolutional autoencoder neural network
IEEE Trans. Big Data
(2017)Self-evolving trading strategy integrating internet of things and big data
IEEE Internet Things J.
(2017)A view of cloud computing
Commun. ACM
(2010)- Cisco Global Cloud Index: Forecast and Methodology, 2015–2020,...
Hedera: dynamic flow scheduling for data center networks
ACM Proc. Netw. Syst. Des. Implementation
(2010)Torus-topology data center network based on optical packet/agile circuit switching with intelligent flow management
IEEE/OSA J. Lightwave Technol.
(2015)Energy-Aware data allocation for mobile cloud systems
IEEE Syst. J.
(2014)
Fog computing, mobile edge computing, cloudlets-whichone?
SoftNet
Maximizing quality of experience through context-aware mobile application scheduling in cloudlet infrastructure
Sotfw. Pract. Exp.
Fog Networking: An Overview on Research Opportunities
Cited by (55)
A methodology to identify identical single-board computers based on hardware behavior fingerprinting
2023, Journal of Network and Computer ApplicationsIKH-EFT: An improved method of workflow scheduling using the krill herd algorithm in the fog-cloud environment
2023, Sustainable Computing: Informatics and SystemsCitation Excerpt :Then, the IKH-EFT algorithm is presented in detail. Fig. 1 shows the architecture of the IKH-EFT method, involving three layers: the terminal layer, the fog layer, and the cloud layer [55,56]. The terminal layer includes smart devices (IoT devices such as home appliances, sensors, smartphones, and smart automobiles).
Deadline-aware and energy-efficient IoT task scheduling in fog computing systems: A semi-greedy approach
2022, Journal of Network and Computer ApplicationsDevice and network driven cellular networks architecture and mobility management technique for fog computing-based mobile communication system
2022, Journal of Network and Computer ApplicationsCitation Excerpt :The simulation results and performance comparison is available in Section 6, the conclusion of work is highlighted in Section 7, thereafter the listed references in the last section. Fog computing association within the cellular network is a demanding area in the research community (Peng et al., 2018). It is widely used in a modern network, 5GB and cellular networks (Liang and Wu, 2020; Zhan et al., 2019).
A new RFID ultra-lightweight authentication protocol for medical privacy protection in smart living
2022, Computer CommunicationsCitation Excerpt :Due to its large capacity and strong computing capability, the cloud server stores some data to facilitate mutual authentication between tags and readers. In medical monitoring systems, attackers often expose users’ privacy through illegal channels [35,36]. As shown in Fig. 3, in medical monitoring systems, the communication between the cloud server and the reader is unreliable because a third party manages the former.
Limei Peng received the M.S. and Ph.D. degrees from the Chonbuk National University,South Korea, in 2006 and 2010, respectively. In 2011, she was a Research Professor with Grid Middleware Research Center, Korea Advanced Institute of Science and Technology, South Korea. She has been an Associate Professor with the School of Electronic and Information Engineering, Soochow University, China, for more than two years. She has been an Assistant Professor with the Department of Industrial Engineering, Ajou University, South Korea. She is now an assistant professor with the School of Computer and Engineering, Kyungpook National University, South Korea. Her research interests fall in optical communication networks and protocols, datacenter networks, software de_ned networks, and cloud computing networks.
Ahmad R. Dhaini is currently an assistant professor of Computer Science at the American University of Beirut (AUB), and a visiting assistant professor at University of Waterloo, while on leave from AUB. He received his B.Sc. in computer science from AUB in 2004; his M.Sc. degree in electrical and computer engineering from Concordia University, Canada in 2006 with a best thesis award nomination. In 2006–2007, he was a software analyst and consultant at TEKSystems, Canada; and in 2007–2008, a software designer at Ericsson, Canada. He obtained his Ph.D. in electrical and computer engineering from University of Waterloo, Canada in 2011, and was granted several awards such as the Ontario Graduate Scholarship in Science and Technology (OGSST), and other various teaching and research awards at University of Waterloo. In 2011–2012, he worked as a research associate at University of Waterloo, Canada, and as a consultant at KAUST, Saudi Arabia. In 2012–2014, he was a postdoctoral scholar at Stanford University, working in the Photonics and Networking Research Laboratory (PNRL), after being awarded the prestigious NSERC postdoctoral fellowship. He also completed the Stanford Ignite program for entrepreneurship and innovation, which teaches scientists how to convert an idea into a business. Dr. Dhaini is a co-inventor of two US patents. He has also authored/coauthored more than 40 highly cited research articles in top IEEE journals and conferences. He is a reviewer for NSF, NSERC, and several US universities’ internal grants. He also serves as Editor for Springer’s Photonics Networks Communications journal, and reviewer and technical program committee (TPC) member for several major IEEE journals and conferences. His research interests cover several themes of integrated networks such as fiber-wireless (FiWi) broadband access networks, mission-critical networks, green communications, edge computing, and software-defined networking. He has also co-PIed several projects related to biotechnology, more specifically in the areas of mobile health and medical image analysis. Pin-Han Ho received the B.Sc. and M.Sc. degrees at the Electrical Engineering Department, National Taiwan University, in 1993 and 1995, respectively, and the Ph.D. degree from Queens University, Kingston, Canada, in 2002. He is currently an Full Professor in the Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, ON, Canada. He is the author/coauthor of more than 150 refereed technical papers, several book chapters, and the coauthor of a book on optical networking and survivability. His current research interests cover a wide range of topics in broadband wired and wireless communication networks, including survivable network design, fiber-wireless network integration, edge computing, and network security. Dr. Ho is the recipient of the Distinguished Research Excellent Award in the Electrical and Computer Engineering Department of University of Waterloo, Early Researcher Award (Premier Research Excellence Award) in 2005, the Best Paper Award in International Symposium on Performance Evaluation of Computer and Telecommunication Systems in 2002, International Conference on Communication Optical Networking Symposium in 2005, and International Conference on Communication Security and Wireless Communications symposium in 2007, and the Outstanding Paper Award in International Conference on High Performance Switching and Routing in 2002
Pin-Han Ho received the B.Sc. and M.Sc. degrees at the Electrical Engineering Department, National Taiwan University, in 1993 and 1995, respectively, and the Ph.D. degree from Queens University, Kingston, Canada, in 2002. He is currently an Full Professor in the Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, ON, Canada. He is the author/coauthor of more than 150 refereed technical papers, several book chapters, and the coauthor of a book on optical networking and survivability. His current research interests cover a wide range of topics in broadband wired and wireless communication networks, including survivable network design, fiber-wireless network integration, edge computing, and network security. Dr. Ho is the recipient of the Distinguished Research Excellent Award in the Electrical and Computer Engineering Department of University of Waterloo, Early Researcher Award (Premier Research Excellence Award) in 2005, the Best Paper Award in International Symposium on Performance Evaluation of Computer and Telecommunication Systems in 2002, International Conference on Communication Optical Networking Symposium in 2005, and International Conference on Communication Security and Wireless Communications symposium in 2007, and the Outstanding Paper Award in International Conference on High Performance Switching and Routing in 2002.