A Survey on IoT Communication and Computation

Frameworks: An Industrial Perspective

Shahab Tayeb1*, Shahram Latifi1, Yoohwan Kim2
1
Department of Electrical and Computer Engineering
2
Department of Computer Science
University of Nevada, Las Vegas
*
shahab.tayeb@unlv.edu

Abstract—This paper surveys fog computing and embedded computational needs, however, is a key issue [1]. With
systems platforms as the building blocks of Internet of Things pervasive computing being the first to revolutionize how and
(IoT). Many concepts around IoT architectures, with various where computation occurs, many vendors introduced their
examples, are also discussed. This paper reviews a high-level products aligned with this notion of ubiquitous computing. For
conceptual layered architecture for IoT from a computational instance, Microsoft Azure became available late 2008 and IBM
perspective. The architecture incorporates fog computing to designed the SmartCloud framework to offer cloud computing
address several issues associated with cloud computing; however, services. Newer paradigms such as fog computing, mist
it is never a binary decision between fog and cloud. Many of the computing, are being devised to extend cloud computing
world’s physical objects are being embedded with sensors and
services to the edge of the network [2] with IOx as one
actuators, tied by communication infrastructures, and managed
by computational algorithms. IoT sensor networks and
prominent example of such an implementation [3]. There have
embedded systems connecting smart objects are revolutionizing been global efforts to standardize the different aspects of IoT
how we approach our daily lives, health care, energy, and and its real-life applications. As an instance, Web of Things
transportation. Such computational needs are addressed with an (WoT) has been used to describe approaches to facilitate
array of various models and frameworks. In an attempt to services offered at Open Systems Interconnection model
consolidate the use of these models, this paper reviews the state- (OSI)’s application layer [4]. IoT not only has the support of
of-the-art research in IoT, cloud computing, and fog computing. providers and business-end giants such as IBM, HP, Intel,
Microsoft, and Cisco but also is fed by innovative services or
Keywords— Cloud Computing; Embedded Systems; Fog the so-called pillars of today’s consumer-oriented Internet;
Computing; IPv6; Smart Objects namely, Apple, Google, and Amazon.
IoT has gained multidisciplinary interest among economists
I. INTRODUCTION at the World Economic Forum (January 2015), collaborating
Internet of Things has recently become a media with Accenture, to identify the potential of connected services
“buzzword” in the realm of Information Technology. Some under the umbrella of Industrial Internet of Things (IIoT) [5].
believe that we are approaching IoT increasingly faster while The electronic environment surrounding us should sense our
others think that we have been in the era of IoT since the presence and respond accordingly, leading to the idea of
beginning of the 21st century. The exact origin of the term ambient intelligence (Aml). Dozens of hardware platforms of
“IoT” and its initial meaning are unknown. The IoT was embedded systems are gaining popularity as prototypes of IoT,
probably coined by Kevin Ashton, co-founder of Auto-ID as part of the Do It Yourself (DIY) projects. The most popular
Center at the Massachusetts Institute of Technology (MIT), as of such embedded-systems platforms and their characteristics
the title of a presentation in 1999. He linked the idea of radio are outlined in section II of this article. Manufacturers
frequency identification (RFID) to the physical world, approach IoT with Machine-to-Machine (M2M), Machine-to-
highlighting people’s roles in data generation leading to a new People (M2P), and People-to-People (P2P) communications.
generation of Internet. Some vendors prefer to name IoT M2M is mainly being utilized to implement “smart factories”;
differently with the prominent example of Cisco Systems, i.e. using IP networks to inter-connect their physical
Accenture, and IBM calling it Internet of Everything (IoE), infrastructure with sensors with added capabilities such as
Internet of Me (IoM), and Smarter Planet, respectively. Sensors analytics and monitoring using technologies such as RFID.
and actuators build the foundation of IoT and with millions of M2P is used to capture and analyze consumer data to be used
sensors in place, the amount of generated data will rise in designing products and services such as mobile marketing to
considerably and hence, computational needs are an ever- push the manufacture-consumer relationship as close to the
growing concern. In a recent IEEE Talks Big Data with Chris consumer as possible. P2P utilizes converged network services
Miyachi, chair of IEEE Cloud Computing, “processing and such as real-time video collaboration tools with Bring Your
analysis” were highlighted as the main chokepoints for Big Own Device (BYOD) capabilities [6]. IoT should preferably be
Data, IoT, and Cyber Physical Systems (CPS) in general analyzed in the context of each specific industry to make sure
because of “the vast amounts of unstructured data that we the discussion fits its application. Proliferation of IoT relies on
currently have in storage”. Where to address such a comprehensive analysis of its limitations; for example, in
This work is supported in part by Doctoral Graduate Research Assistantship
from UNLV Graduate College and in part by the NSF award #EPS-IIA-
1301726 (EPSCoR NEXUS).

978-1-5090-4228-9/17/$31.00 ©2017 IEEE

terms of security, privacy, and cost effectiveness. This paper the realm of smart home and a short description of each are as
surveys many concepts around IoT, particularly, from a follows: V. Sandeep, et al. [11] proposed a remotely-controlled
computation perspective and addresses common challenges automation system for home electrical appliances using cloud.
with IoT architectures. This paper is organized as follows: This was done using a secondary machine to do the thinking.
Section 2 presents IoT’s current platforms and technologies, N. Agrawal, et al. [12] introduced a drip irrigation system as a
from prototype embedded systems to communications and proof of concept for residential purposes which can be applied
backbone technologies to IoT applications. Section 3 discusses to agricultural fields and gardens, simply by sending an email
a layered IoT architecture and provides an in-depth review of to the system. Raspberry PI has also been used for modeling
each layer. Section 4 discusses research challenges. cloud computing clusters in [13] and [14] or construction of
virtual supercomputers [15]. Raspberry PI is also used to target
II. IOT PLATFORMS AND TECHNOLOGIES K-12 education and has a wide support for school projects,
making it an integral part of IoT’s future. Partitioning IoT
A. Embedded Systems Platforms for IoT Prototypes devices into communities can help better management and
organization of information flow. Pirouz et al. [16] proposed an
Y. Li et al. [7] introduced a theoretical framework, optimized distance-based metric to prioritize similar nodes.
demonstrating the growing need for IoT application in different
contexts and industries. There has been an industrial trend in
popularizing programmability of hardware components. This is B. IoT Communications Technologies
mainly because consumers drive the need for products, and IoT networks utilize many traditional networking protocols
migration to IoT is impossible without involving different operating at physical and data link layers as well as some
social groups such as students. To make IoT a reality, a variety newer protocols and standards that are mainly designed for IoT
of programmable hardware platforms have gained popularity. applications. 802.15 wireless personal area networks (WPANs)
Some of the most popular platforms, their features, and standards are used for short range communication, typically
connectivity solutions are represented in Fig. 1. between 1m and 100m) such as 802.15.1 (Bluetooth
compatible) and 802.15.4 (ZigBee compatible). The former is a
point-to-point technology while the latter operates based on a
star logical topology, dividing the peer-to-peer topography into
cluster formations; thus, supporting broadcast in addition to
unicast transmissions. Other short-range wireless protocols
supported by many IoT objects are Near-Field Communication
(NFC) which supports proximity of centimeters rather than
meters. Hybrid standards such as IPv6 over Low power
Wireless Personal Area Networks (6LoWPAN) offer low-
power IoT objects with smaller encapsulation by compressing
the header. Even though 6LoWPAN is designed for IPv6
nodes, efforts have been made to provide interoperability of
6LoWPAN and its application to IPv4 networks as presented in
[17]. Some researchers are introducing more efficient
alternatives to these protocols. For instance, S.H. Kim et al.
[18] introduced a UPnP-ZigBee internetworking architecture
where a UPnP gateway mirrors a ZigBee topology using UPnP
proxies. 802.11 standards are mainly used for communications
up to 32m. For longer distances, cellular networks are mainly
used. Many such wireless technologies and their performance
in different fields are analyzed in the literature [19]. Web
search popularity of paradigms used in this paper are illustrated
in Fig. 2. These numbers were obtained using Google search
trends. A value of 100 is the peak popularity for the term. Both
“Cloud Computing” and “Raspberry Pi” term popularities are
Fig. 1. System-on-Chips (SoC) for IoT Prototypes (sorted alphabetically) on the rise with slight fluctuations.

Introduced in February 2012, Raspberry PI is by far the

most popular and cheapest platforms and is a modular and
flexible tool for not only real world applications but also
educational purposes. C. Edwards, et al. [8] were one of the
very first publications introducing an array of Raspberry PI
applications. S. Joardar, et al. [9] developed a biometric system
for human subject recognition based on the palm vein pattern
using Raspberry PI. Similarly, S. Sivaranjani, et al. [10]
implements a model for extraction of fingerprint and footprint Fig. 2. Web search trends for terms: Fog Computing, Internet of Things,
in newborn babies. Some novel applications of Raspberry Pi in Raspberry Pi, and Cloud Computing since 2007
 Some of the main IoT communication standards and used to identify the potential security vulnerabilities. Smart
proprietary protocols are outlined in Fig. 3. objects are categorized based on their attributes. The reason for
defining different classes is for implementation purposes,
either in the form of 3-bit or 7-bit flag bits. The former is a
similar implementation as Type of Service (ToS) bits in IPv4
or differentiated services (DiffServ) bits in IPv6 header for
Quality of Service (QoS) purposes while the latter resembles
the flag bits in IPv4 and IPv6 headers. Some examples of
existing attack types and their potential reflection to IoT
networks are a) DoS attacks rendering a device unusable
through exhaustion of target’s resources - IoT end-devices
have more limited resources, e.g. CPU & RAM; b) DDoS
where DoS is initiated by thousands of zombies - 50 billion
devices in IoT can become zombies and the same 50 billion are
potential victims; c) Eavesdropping as a reconnaissance attack
- more data for IoT leads to a higher probability of
reconnaissance gaps; d) Sybil or subversion of reputation
systems by forged identities in peer-to-peer networks -
Wireless Sensor Networks are the main target for Sybil attacks;
and e) Black hole where packets are dropped on intermediary
devices - limited resources on IoT sensors as easy targets.
For the security of lower OSI layers, concepts such as
collision domain and broadcast domain were formed with the
introduction of networking devices such as Ethernet switches
Fig. 3. Enabling Wireless Technologies and routers. These intermediary nodes can understand source
and destination addresses and make CAM and routing tables
for their lookup processes. Such devices replaced hubs, which
C. IoT Backbone Technologies
flooded the ingress packet to all operational ports. This does
A brief description of the underlying backbone protocols not address the security concerns in full but is the fundamental
driving IoT architecture are given here for completeness. concept for more advanced features such as Dynamic Address
Additional information on any of the following can be obtained Resolution Protocol (ARP) Inspection (DAI) that only allows
by referring to the particular standards. Internet Protocol, valid ARP requests and responses to pass through, dropping
versions 4 and 6, support almost all implementations of IoT at invalid IP-to-MAC address bindings. Furthermore, based on
layer 3 of the OSI model. IPv6 has taken over IPv4 because of CAM entries, security features such as port security are
exhaustion of IPv4 address range. However, because of the deployed, adding another layer of security at bottom OSI
existing IPv4 nodes and devices, interoperability is provided by layers. Several short-range wireless protocols also support
other layer protocols. 6LoWPAN defines smaller encapsulation communications. Standards such as 6LoWPAN offer low-
by header compression over 802.15.4 networks. Objects can power IoT objects with smaller encapsulation by compressing
become part of IoT if they meet some basic requirements. the header. To be able to model CPS attacks, we first need to
These requirements are: a) connectivity: usually small categorize them. One of the common approaches is to divide
bandwidth wires or wireless connections, b) power: usually no attacks into four distinct categories. A security system can be
opportunity for charging or replacing battery, c) processor: proposed based on the security metrics (Equation 1).
low-powered cost-effective microprocessors for computing, d)
storage: usually limited local storage, e) security: generation S = a * PA + β * CA + γ * I + θ * C (1)
and transmission of sensitive and regulated information, f)
reliability: operability for extended periods of time in harsh where S is the overall security of the system; α, β, γ, and θ
environments and outdoors, and g) sensing: making objects are coefficients that show the relative importance of the
measuring and generating data into distinguishable and security parameters for Physical Availability (PA), Cyber
addressable supporting queries. Availability (CA), Integrity (I), and Confidentiality (C),
respectively. To generalize Equation 1, one can take the
D. IPv6 and Security for IoT number of PA attacks to be between [1,n], the number of CA
attacks to be between [1,m], the number of I attacks to be
IPv6 is inherently more secure than IPv4. This is reflected
between [1,p] and the number of C attacks to be between [1,k].
in its reformed header, using extended options [20].
Thus, the sum of all possible attacks can be introduced in form
Additionally, IPv6 features such as Link-Local Addresses
of Equation 2.
(LLAs) are secure by design because they are not available to
outside of the link, so no malicious traffic can be sent remotely.
Similarly, replacing broadcast messages with anycast packets
(2)
diminishes flooding attacks, which are the basis for various
Denial of Service (DoS) and Distributed Denial of Service
(DDoS) attacks. Vulnerabilities and existing attack models are
E. IoT Applications Additionally, these objects are mostly insecure because of their
IoT encompasses various aspects of our daily lives. Smart inherent limited resources and power. As a result of utilizing a
home solutions, connected-body sensors, smart city, smart wide array of heterogeneous and often unreliable smart objects
transportation, and enterprise applications are a few of these in building these networks, there is a need for a reliable design
aspects. In the context of smart city applications of IoT, some model capable of supporting high-goodput applications. These
examples of related projects are: a) Sintelur waste management smart objects encompass small wearable technologies which
which determines the filling level of various types of are limited in power to large-scale deployment of rich-
recyclable waste; namely, glass, paper, or aluminum cans using application embedded systems in smart city architecture. From
M2M communication on GPRS. The city government has a a security standpoint, the design of this layer should not only
dashboard to easily manage the recycling procedure. This address security concerns but also provide on-demand security
system also computes the best route to reduce CO2 emissions guidelines for the next generation of small and large embedded
[21]; b) Thingful, a discoverability engine based on twitter sensor systems. App Execution Platform (AEP) is one of the
profiles categorizes geographical index of locations, virtualization paradigms for interaction and communication of
ownerships, and reason for public and private resources. This such objects [23].
project was the people’s choice winner of 2013 Internet of
Things awards [21]; and c) Streetline utilizes sensors to guide B. Fog Computing Layer
drivers to available parking spots and thus, reducing city The fog computing layer acts as the interface between the
traffic, noise pollution, and CO2 emissions [21]. Other smart objects layer and the cloud layer. Smart objects are
applications are smart and connected health care [22]. physical devices generating different types of information
using a variety of sensors and with connectivity to the
III. IOT COMPUTING FRAMEWORK backbone, they are capable of transmitting their data to other
nodes. To accommodate for such a heterogeneous list of smart
The proposed framework follows PPDIOO network objects, this multi-modal layer supports many different
lifecycle as outlined in Fig. 4a. This will ensure a design that protocols and is the aggregation point for different packet
meets the goals of IoT and will lead to a comprehensive design types. Modularity and security are the main characteristics of
following the proactive management of operate and optimize this layer. Modularity supports scalability through seamless
stages. Continuous improvement of the designed framework is integration of new components into the infrastructure. Security
possible through proactive management of the auditing nodes. services ensure a reliable network and cover the limited power
This will be added as the last component of the lifecycle to and computational resources of the smart objects. A proposed
provide logging services and real-time detection of behavior ubiquitous-computing middleware for smart objects is
anomalies. From a computational perspective, IoT can be presented in [24] where the authors use metadata for
illustrated in form of a three-layer hierarchical design (Fig. 4b). interoperability of services. The advantage of this middleware
The proposed model is based on a secure framework is that it is transparent to the smart object, making it usable for
introduced by Cisco. The underlying infrastructure of IoT is different applications of IoT and a variety of CPS.
comprised of millions of smart objects. These layers are Virtualization techniques at this layer can lead to cumulative
described below: virtual storage systems which, in contrast to traditional storage
units, provide ample storage opportunities and are a one-to-
near service. This can result in lower latency and delay for
storage services. L.F. Zeng, et al. [25] introduced smart object-
based storage system using machine learning and pattern
recognition. Another characteristic of this layer is its capability
to discover services and route the traffic to the shortest, nearest
gateway or storage unit. Local and global service discovery
mechanisms, although independent, can be operationally based
on the same architecture. S. Cirani et al. [26] proposed a
framework for this purpose. Fog computing is extending cloud
computing services to the network edge.
Fog computing is criticized in some literature because of its
insecurity. I. Stojmenovic et al. [27] elaborate the advantages
Fig. 4. a. PPDIOO lifecycle [recreated from 22]; b. three-layer framework of fog computing, building their discussion towards
for IoT from a computing perspective highlighting the security and privacy concerns of such
implementations. It should be noted that similar to other
A. Smart Objects Layer frameworks, fog computing is not secure by design and
additional security configurations and planning is needed to
S mart objects are embedded systems, actuators, and ensure security and privacy of such systems. Fog computing is
sensors with different CPU, memory, and Operating Systems. a relatively new term, introduced in 2012. Instances of fog
Most of such devices are single-function devices with network computing in IoT applications are also analyzed in some
connectivity. More often than not, these objects are remote publications. Fog computing in the healthcare system and its
and/or located in inaccessible locations. Hence, neither advantages, mainly in terms of lowering the latency, is
modification of the object nor reconfiguring prove practical.
introduced in [28]. A mathematical modeling of fog D. Research Challenges
computing, proving that it outperforms cloud computing is It is widely accepted that the underlying IoT infrastructure,
given in [29]. The reasons behind using fog computing are its technologies, and applications are broad, but still in their
twofold: firstly, pushing computation from the cloud to the infancy [42]. Lack of proper standardization is a major barrier
network, and secondly, adding security by pushing ahead of IoT research. Some standardization organizations
computations to the network edge. Y.N. Krishnan, et al. [30] such as ITU-T have started standard initiatives for this purpose;
introduced a virtual AppStore on network intermediary devices for example, IoT-GSI, or IoT Study Group 17, which focuses
which supports the former reason. Y Wang et al. [31] on security and privacy of tag-based applications. ITU-T
reviewed motivations behind replacing cloud computing with technology watch report: Ubiquitous sensors network
fog computing and concluded that further extensive research is published in February 2008 is another reference document on
required to justify this replacement. However, in an earlier this topic [43]. The ever-growing number of devices and
work, [32] considered the reasons behind using fog computing services connected to IoT makes it difficult to propose a
and proposed a policy-based resource and security coherent framework but with the added modularity introduced
management in support of fog computing. In another work by the proposed hierarchical framework, scalability and future
[33], authors strongly consider fog computing to be the natural growth is supported. Even so, with a large number of physical
platform for IoT and conclude with the limitations of some of objects, scalability might be intertwined with data transfer and
the existing technologies that should support fog computing. processing [44]. As a result of so many heterogeneous smart
objects, a great deal of streaming and real-time data will be
C. Cloud Layer generated and this will give rise to expected and unexpected
Cloud layer acts as the backbone for IoT, similar to the problems [45]. All the generated data might not prove useful;
traditional network backbone but with some differences. One in which case, a decision should be made whether to either
of the main differences is the flexibility and adaptability of the store the data or discard it. Storing it might prove useful in the
cloud layer. Unique protocols with different packet sizes need future; however, this task could prove tedious [46].
to be transmitted or computed at this layer. Security of this
layer is a critical task as this layer, by nature, is prone to IV. CONCLUSION
attacks such as Replay attacks, Man-in-the-Middle, and
Spoofing. This layer does not have a rigid edge and hence the This paper reviewed various IoT concepts and the existing
name cloud. Advances in cloud computing and cloud concepts computation techniques from an industrial viewpoint. Initially,
make it hard to have a consolidated definition for this layer. the hardware, protocols, and some applications of IoT were
B.P. Rimal, et al. [34] proposes a comprehensive taxonomy of surveyed. Next, a layered framework for computation in IoT
cloud computing and maps existing projects such as Amazon was proposed and the need for each layer was addressed
to their proposed model. Mobile computing architectures and separately. Computational approaches such as fog and cloud
cloudlets are possible implementation models of the cloud computing were mapped in the proposed layered framework. A
layer. Several proof-of-concept prototypes have been proposed mathematical model for the security of these paradigms was
in the literature. M. Satyanarayanan, et al. [35] applied virtual proposed, incorporating inherent IPv6 security measures.
machine (VM) to mobile users while K. Kumar [36] addressed
the energy aspect of VM-based cloudlets. Cloud layer REFERENCES
implementations are usually divided into three main categories: [1] IEEE Talks Big Data Q&A Series with C. Miyachi, [online, accessed:
public, private, and hybrid clouds. The main reason is, as May 2016] http://bigdata.ieee.org/publications/ieee-talks-big-data
proposed in [37], cloud solutions can be offered as [2] J.S. Preden, K. Tammemae, A. Jantsch, M. Leier, A. Riid, E. Calis, "The
infrastructure-as-a-service (IaaS) meaning that private and Benefits of Self-Awareness and Attention in Fog and Mist Computing,”
hybrid solutions are a specific type of public clouds whose Computer, vol.48, no.7, pp. 37-45, July 2015.
services are virtually leased to users and organizations. This [3] A.V. Dastjerdi, H. Gupta, R. Calheiros, S. Ghosh, R. Buyya, “Fog
Computing: Principles, Architectures, and Applications,” Internet of
does not necessarily affect the actual underlying infrastructure Things: Principles and Paradigms, Morgan Kaufmann, Burlington,
of the cloud or its services and is mainly a discussion of who Massachusetts, USA, 2016.
the target user is. Proposed research paths and various [4] D. Guinard, V. Trifa and E. Wilde, "A resource oriented architecture for
implementations of cloud computing alongside their the Web of Things," Internet of Things, 2010, Tokyo, 2010, pp. 1-8.
disadvantages and concerns are discussed in detail in [38]. For [5] A.-R. Sadeghi, C. Wachsmann, M. Waidner, "Security and privacy
a more in-depth analysis of cloud computing security and challenges in industrial Internet of Things,” Design Automation Conf.
privacy, [39] analyses such issues and provides the reader with (DAC), 2015 52nd ACM/EDAC/IEEE, pp. 1-12, June 2015.
solutions to overcome them. Cloud layer has the potential to be [6] G. Tanganelli, E. Mingozzi, C. Vallati, C. Cicconetti, "A distributed
used for storage purposes with flexibilities such as append, architecture for discovery and access in the internet of things,” IEEE
INFOCOM WKSHPS, pp. 45-46, 14-19 April 2013.
delete, and update; however, being a new frontier, these are not
[7] Y. Li, M. Hou, H. Liu, and Y. Liu, “Towards a theoretical framework of
error-free environments. An error model of cloud computing strategic decision, supporting capability and information sharing under
architecture using a Byzantine failure is proposed in [40]. the context of Internet of Things,” Inf. Technol. Manage., vol. 13, no. 4,
Several open-source cloud computing systems exist. One of pp. 205–216, 2012.
such systems named Eucalyptus is proposed in [41]. [8] C. Edwards, "Not-so-humble raspberry pi gets big ideas,” Engineering
& Technology, vol.8, no.3, pp. 30-33, April 2013.
[9] S. Joardar, A. Chatterjee and A. Rakshit, "A Real-Time Palm Dorsa
Subcutaneous Vein Pattern Recognition System Using Collaborative
 Representation-Based Classification," in IEEE Trans. on [27] I. Stojmenovic, W. Sheng "The Fog computing paradigm: Scenarios and
Instrumentation and Measurement, vol. 64, no. 4, pp. 959-966, April security issues,” Computer Science and Information Systems (FedCSIS),
2015. doi: 10.1109/TIM.2014.2374713 2014 Federated Conf., pp. 1-8, 7-10 Sept. 2014.
[10] S. Sivaranjani, S. Sumathi, "Implementation of fingerprint and newborn [28] Y. Shi, G. Ding, H. Wang, H.E.Roman, S. Lu, "The fog computing
footprint feature extraction on Raspberry Pi,” Innovations in service for healthcare,” Future Information and Communication
Information, Embedded and Communication Systems (ICIIECS), 2015 Technologies for Ubiquitous HealthCare (Ubi-HealthTech), 2015 2nd
Int’l Conf., pp. 1-6, 19-20 March 2015. Int’l Symposium on, pp. 1-5, 28-30 May 2015.
[11] V. Sandeep, K.L. Gopal, S. Naveen, A. Amudhan, L.S. Kumar, [29] S. Sarkar, S. Chatterjee, S. Misra, "Assessment of the Suitability of Fog
"Globally accessible machine automation using Raspberry pi based on Computing in the Context of Internet of Things,” in IEEE Trans. on
Internet of Things,” Advances in Computing, Communications and Cloud Computing, no.99.
Informatics (ICACCI), Int’l Conf., pp. 1144-1147, 10-13 Aug. 2015. [30] Y.N. Krishnan, C.N. Bhagwat, A.P. Utpat, "Fog computing — Network
[12] N. Agrawal, S. Singhal, "Smart drip irrigation system using raspberry pi based cloud computing,” Electronics and Communication Systems
and arduino,” Computing, Communication & Automation (ICCCA), (ICECS), 2015 2nd Int’l Conf., pp. 250-251, 26-27 Feb. 2015.
2015 Int’l Conf., pp. 928-932, 15-16 May 2015. [31] Y. Wang, T. Uehara, R. Sasaki, "Fog Computing: Issues and Challenges
[13] P.T. Fung, D.R. White, S. Jouet, J. Singer, D.P. Pezaros, "The Glasgow in Security and Forensics,” Computer Software and Applications Conf.
Raspberry Pi Cloud: A Scale Model for Cloud Computing (COMPSAC), 2015 IEEE 39th Annual, vol.3, pp. 53-59, 1-5 July 2015.
Infrastructures,” Distributed Computing Systems Workshops (ICDCSW), [32] C. Dsouza, G.-J. Ahn, M. Taguinod, "Policy-driven security
2013 IEEE 33rd Int’l Conf., pp. 108-112, 2013. management for fog computing: Preliminary framework and a case
[14] P. Abrahamsson, S. Helmer, N. Phaphoom, L. Nicolodi, N. Preda, L. study,” Information Reuse and Integration (IRI), 2014 IEEE 15th Int’l
Miori, M. Angriman, J. Rikkila, W. Xiaofeng, K. Hamily, S. Bugoloni, Conf., pp. 16-23, 13-15 Aug. 2014. doi: 10.1109/IRI.2014.7051866
"Affordable and Energy-Efficient Cloud Computing Clusters: The [33] M. Yannuzzi, R. Milito, R. Serral-Gracia, D. Montero, Nemirovsky, M.,
Bolzano Raspberry Pi Cloud Cluster Experiment,” Cloud Computing "Key ingredients in an IoT recipe: Fog Computing, Cloud computing,
Technology and Science (CloudCom), 2013 IEEE 5th Int’l Conf., vol.2, and more Fog Computing,” Computer Aided Modeling and Design of
pp. 170-175, 2-5 Dec. 2013. doi: 10.1109/CloudCom.2013.121 Communication Links and Networks (CAMAD), 2014 IEEE 19th Int’l
[15] P. Turton, T.F. Turton, "PiBrain — A cost-effective supercomputer for Workshop on, pp. 325-329, 1-3 Dec. 2014.
educational use,” Engineering and Technology (BICET 2014), 5th [34] B.P. Rimal, E. Choi, I. Lumb, "A Taxonomy and Survey of Cloud
Brunei Int’l Conf., pp. 1-4, 1-3 Nov. 2014. doi: 10.1049/cp.2014.1121 Computing Systems,” INC, IMS and IDC, 2009. NCM '09. Fifth Int’l
[16] Matin Pirouz, and Justin Zhan. "Optimized relativity search: node Joint Conf., pp. 44-51, 25-27 Aug. 2009. doi: 10.1109/NCM.2009.218
reduction in personalized page rank estimation for large graphs." Journal [35] M. Satyanarayanan, P. Bahl, R. Caceres, N. Davies, "The Case for VM-
of Big Data 3.1 (2016): 12. doi: 10.1186/s40537-016-0047-2 Based Cloudlets in Mobile Computing,” in IEEE Pervasive Computing,
[17] C.Y Yum, Y.S. Beun, S. Kang, Y.R. Lee, J.S. Song, "Methods to use vol.8, no.4, pp. 14-23, Oct.-Dec. 2009. doi: 10.1109/MPRV.2009.82
6LoWPAN in IPv4 network,” Advanced Communication Technology, [36] K. Kumar, Y.-H. Lu, "Cloud Computing for Mobile Users: Can
The 9th Int’l Conf., vol.2, pp. 969-972, 12-14 Feb. 2007. Offloading Computation Save Energy?,” in Computer, vol.43, no.4, pp.
[18] S.H. Kim, J.S. Kang, H.S. Park, D. Kim, Y.J. Kim, "UPnP-ZigBee 51-56, April 2010. doi: 10.1109/MC.2010.98
internetworking architecture mirroring a multi-hop ZigBee network [37] B. Sotomayor, S. Montero, S. Ruben I.M. Llorente, I. Foster, "Virtual
topology,” in IEEE Trans. on Consumer Electronics, vol.55, no.3, pp. Infrastructure Management in Private and Hybrid Clouds,” in IEEE
1286-1294, August 2009. doi: 10.1109/TCE.2009.5277989 Internet Computing, vol.13, no.5, pp. 14-22, Sept.-Oct. 2009.
[19] A. Willig, K. Matheus, A. Wolisz, "Wireless Technology in Industrial [38] M.A. Vouk, "Cloud computing - Issues, research and implementations,”
Networks,” in Proceedings of the IEEE, vol.93, no.6, pp. 1130-1151, Information Technology Interfaces. ITI 2008, 30th Int’l Conf., pp. 31-40,
June 2005. doi: 10.1109/JPROC.2005.849717 23-26 June 2008.
[20] P. Nikander, A. Gurtov, T. R. Henderson, "Host Identity Protocol (HIP): [39] T. Dillon, C. Wu, E. Chang, "Cloud Computing: Issues and Challenges,”
Connectivity, Mobility, Multi-Homing, Security, and Privacy over IPv4 Advanced Information Networking and Applications (AINA), 2010 24th
and IPv6 Networks," in IEEE Communications Surveys & Tutorials, vol. IEEE Int’l Conf., pp. 27-33, 20-23 April 2010.
12, no. 2, pp. 186-204, Second Quarter 2010.
[40] C. Wang, Q. Wang, K. Ren, W. Lou, "Ensuring data storage security in
[21] Nominated Projects for 4th Annual Internet of Things Awards [Last Cloud Computing,” Quality of Service, 2009. IWQoS. 17th Int’l
accessed: October 2015] Available: http://postscapes.com/internet-of- Workshop on, pp. 1-9, 13-15 July 2009.
things-award/2014/
[41] D. Nurmi, R. Wolski, C. Grzegorczyk, G. Obertelli, S. Soman, L.
[22] M. Hassanalieragh et al., "Health Monitoring and Management Using Youseff, D. Zagorodnov, "The Eucalyptus Open-Source Cloud-
Internet-of-Things (IoT) Sensing with Cloud-Based Processing: Computing System,” Cluster Computing and the Grid, CCGRID '09. 9th
Opportunities and Challenges," 2015 IEEE International Conference on IEEE/ACM Int’l Symposium on, pp. 124-131, 18-21 May 2009. doi:
Services Computing, New York, NY, 2015, pp. 285-292. doi: 10.1109/CCGRID.2009.93
10.1109/SCC.2015.47
[42] L. Xu, “Enterprise Systems: State-of-the-art and future trends,” in IEEE
[23] M. Stecca, C. Moiso, M. Fornasa, P. Baglietto and M. Maresca, "A Trans. Ind. Informat., vol. 7, no. 4, pp. 630–640, Nov. 2011.
Platform for Smart Object Virtualization and Composition," in IEEE
Internet of Things Journal, vol. 2, no. 6, pp. 604-613, Dec. 2015. [43] Ubiquitous Sensor Networks (USN), ITU-T Technology Watch Briefing
Report Series, No. 4 (February 2008).
[24] EunJung Ko, J. Kang, J. Park, "A middleware for smart object in
[44] D. Miorandi, S. Sicari, F. De Pellegrini, I. Chlamtac, “Internet of things:
ubiquitous computing environment," Computing Technology and
Information Management, 2012 Int’l Conf on, Seoul, pp. 400-403, 2012. Vision, applications and research challenges,” Ad Hoc Netw., vol. 10,
no. 7, pp. 1497–1516, 2012.
[25] L.F. Zeng, D. Feng, L.J. Qin, "SOSS: smart object-based storage
[45] S. Wang, Z. Zhang, Z. Ye, X. Wang, X. Lin, S. Chen, “Application of
system,” Machine Learning and Cybernetics, 2004. Proceedings of 2004
Int’l Conf., vol.5, pp. 3263-3266 vol.5, 26-29 Aug. 2004. environmental internet of things on water quality management of urban
scenic river,” Int. J. Sustain.Develop.World Ecol., vol. 20, no. 3, pp.
[26] S. Cirani, L. Davoli, G. Ferrari, R. Leone, P. Medagliani, M. Picone, L. 216–222, 2013.
Veltri, "A Scalable and Self-Configuring Architecture for Service
Discovery in the Internet of Things,” in IEEE Internet of Things [46] C. Perera, A. Zaslavsky, P. Christen, D. Georgakopoulos, "Context
Journal, vol.1, no.5, pp. 508-521, Oct. 2014. Aware Computing for The Internet of Things: A Survey,”
Communications Surveys & Tutorials, IEEE, vol.16, no.1, pp. 414-454,
First Quarter 2014. doi: 10.1109/SURV.2013.042313.00197