Journal of Ambient Intelligence and Humanized Computing

https://doi.org/10.1007/s12652-020-02448-3

ORIGINAL RESEARCH

Cloud of Things: architecture, applications and challenges

Fahd Alhaidari1 · Atta Rahman1 · Rachid Zagrouba1

Received: 5 March 2019 / Accepted: 30 July 2020

© Springer-Verlag GmbH Germany, part of Springer Nature 2020

Abstract
Nowadays, Cloud Computing and Internet of Things (IoT) are amongst the most emerging internet technologies
playing a vital role in our daily lives. In the near future, their usage is expected to be increased exponentially
that is making them the important technologies of the future internet. Cloud Computing provides access to
unlimited capacities and capabilities, such as computation, processing, and storage through a shared group of
resources. That can easily be acquired and dynamically assigned to potential IoT based applications. Because,
IoT devices and applications are rather defcient in terms of these capabilities. This paper is primarily dedicated
to investigating the integration of technologies of Mobile Computing, Cloud Computing and IoT; coined as
Cloud of Things (CoT), its characteristics, architectures, potential challenges with possible solutions and
applications.

Keywords Cloud Computing · Mobile Computing · Cloud of Things · IoT · MCC · CoT issues and
challenges · CoT applications

1 Introduction Kingdom of Saudi Arabia

a new technology called Cloud of Things; it shows
Cloud Computing and Internet of Things (IoT) are two that IoT functions can be improved by mobile and
dif ferent as well as new virtualization technologies. cloud computing technology (Stergiou et al. 2018;
Cloud Computing provides unlimited capabilities, Aazam et al. 2014). That is basically augmenting a
mainly in terms of data storage and data processing technology which is relatively poor in terms of
over the Internet as a service acquired by consumer. capabilities but rich in terms of real-time data to a
While IoT refers to a system that consists of stronger technology in terms of processing
connected devices that can transfer and use data capabilities. Another, integral part of this
over a wireless network to mainly provide utility ser technological chain is com
vices like appliances control etc. These devices could prised of mobile computing that is although
be of many types, starting from home appliances considered as a constituent part of cloud computing
connected to internet to distributed sensors across (termed as mobile cloud computing) yet has its own
any geographical location connected via low earth identity. In terms of capa bilities, the mobile computing
orbit (LEO) satellites (Qu et al. 2017). The IoT layer can be in the middle of IoT and Cloud.
devices are mostly like dumb terminals or think Precisely, if IoT and Cloud are considered as thin and
clients, generating sensory data but unable to pro thick clients, respectively; mobile computing can be
cess it at their own. That is why mainly there is a dire designated as a hybrid client model. So, the mobile
need for associated servers at diferent levels of comput ing is somewhat a middleware in Cloud of
hierarchy like fog, edge and eventually cloud (Dash Things. That is, the network part provides the means
et al. 2019; Devara jan et al. 2019; Jin et al. 2019). of communication and connectively through various
The integration of these two technologies namely links that could be wired and/ or wireless links like Wi-
Cloud Computing and IoT yields Fi, WiMAX, mobile cellular and satellite etc. (Somula
and Sasikala 2018).
Similarly, in (Elazhary 2019; Bittencourt et al. 2018),
🖂 Atta Rahman the authors have precisely distinguished several
[email protected] emerging tech nologies like IoT, edge, fog, mobile
and cloud computing and their possible overlaps as
1
College of Computer Science and Information shown in Fig. 1. It is apparent from this
Technology, Imam Abdulrahman Bin Faisal comprehensive depiction, that all these technolo gies
University, P.O. Box 1982, Dammam,
 somewhat overlap with each other and technological paradigms. Moreover,
generate/produce new research areas and

Vol.:(0123456789) 13
F. Alhaidari et al.

and computing requirements at the bigger scale. IoT,

on the other hand, rather a data generation layer.
For the communication requirements, due to cloud
com puting powers and the availability of high speed
networks, CoT provides an efective and efcient set of
techniques
to connect, control, and monitor remote things from
any where at any time considering real time access
and low cost of communications (Rawahi et al. 2018;
Suciu et al. 2013). For the storage requirements, data
generated by IoT comes with several concerns
including its volume, its structure (mostly non-or
semi-structured), and its rate of generation (Botta
et al. 2016). Thus, there is a need of powerful
storages that can deal with such sort of data in low
cost accompanied by efcient and secure access. CoT
provides such storages derived and managed by
cloud computing layer enabling many services on the
data including archiving, aggregation, integration, and
Fig. 1 Overlapping areas among recent computing paradigms sharing (Zaslavsky et al. 2013).
(Elazhary 2019) CoT provides the required computation power
through the utilization of cloud computing processing
capabilities available through pay per usage model.
according to the diagram, the most enriched, covered The cloud comput ing as a layer of CoT solves many
and technologically sound region is Mobile IoT Cloud concerns involved in IoT computational paradigm
Comput ing. That is the focused area in this paper, such as scalability, real time analysis, and limited
and it is termed as Cloud of Things. power of IoT nodes (Yao et al. 2013; Jiang et al.
Both cloud computing and IoT have their own 2019). Thus, with CoT, smart IoT services can be
capabilities as well as limitations due to the nature of provided due to the computational power available
their architecture and environment. Cloud computing through cloud computing.
focused on providing computing resources available Generally, integrating cloud computing and IoT into
through datacenters and delivered based on a utility a CoT model resolves the limitations of both alone
model. Hence, the communica (Aazam et al. 2014; Rawahi et al. 2018). However,
tion between datacenters and outside environments is CoT as a new paradigm rises many challenges that
limited (Narwane et al. 2019). On the other hand, IoT required attention of researchers to provide solutions
designed to interconnect heterogeneous things for such challenges (Díaz et al. 2016). Example of
together as well as with the outside environment. The such challenges includes Energy Efciency, Resource
main limitation of IoT is its limited computing and Allocation, Identity Management, Service Discovery,
storage resources due to the capacity of things like Location of Data Storage, Security and Privacy,
sensors and actuator (Lee et al. 2010). Communication of unnecessary data, Heterogene ity,
Due to huge volume of data (big data) being Performance, and dealing with Big Data (Rawahi et al.
generated and communicated by massive number of 2018; Alrawahi et al. 2019).
connected devices, people, and things, Cloud of This research intends to provide a state-of-the-art
Things (CoT) as an integra tion of IoT with cloud study on CoT model, application areas along with the
computing becomes a necessary and important challenges that might encounter the deployment of
technology to manage such data according to its this technology. Solutions for these concerns will be
requirements to create more valuable and smart highlighted and pro
services and applications. CoT provides the required posed based on the existing studies as well as based
environment to deal with the overwhelming on authors knowledge in this area.
requirements and specifca tions of the big data In this paper, integration of most recent and
generated by IoT devices and sensors. Such promising technologies namely, Mobile Cloud
environment facilities the communication, storage, Computing (Mobile Computing and Cloud Computing)
 and IoT has been investigated along with their describes the IoT technology. Section 4 describes
characteristics, architec the Cloud of Things in depth, Sect. 5 describes the
tures, potential challenges with their possible solutions issues and possible/ available solutions to them.
and applications. Section 6 provides potential applications of CoT,
Rest of the paper is organized as follows: Sect. 2 is future challenges and their projected solutions, while
ded icated to mobile and cloud computing, Sect. 3 Sect. 7 concludes the paper.

13
Cloud of Things: architecture, applications and challenges

2 Mobile and cloud computing

The advent of mobile devices has added several

features that facilitated users’ lives. It provides them
with real-time com munication services like global
positioning system (GPS), ability to access external
modern services and applications on the internet, and
many other features. However, these mobile devices
have limitations such as battery, storage, processing
and bandwidth. Therefore, a solution to extend
mobile devices capabilities is highly needed. Cloud
comput ing provides mobile devices with several
advantages that help them to overcome their
limitations. Such advantages include extra data
storage, cloud services for data process ing, and
perform all resource-intensive computing on the
could. The introduction of cloud computing made it
possible to use infrastructures, platforms, and resources to the applications. The last layer is the
software provided by cloud providers at a minimum physical layer, which contains switches, routers,
cost. Moreover, cloud com puting provided users with individual servers, and some resources that help the
on-demand resources consump tion. With the application and virtualiza tion layers to perform their
increase of mobile apps and the help of cloud tasks (Qu et al. 2017). Figure 2 shows the architecture
computing for numerous services for mobile users, of cloud computing.
mobile cloud computing is introduced as an Software as a Service (SaaS), Platform as a
integration of cloud computing into the mobile Service (PaaS), and Infrastructure as a Service
environment. (IaaS) are three main service models used to defne
Cloud Computing is a remote server platform used to Cloud services. In SaaS, the users can access Cloud
pro vide computing, storage, applications, and application over the Internet without the need of
hardware services so that users can store and utilize installing and storing the application. The advan
large amounts of data over the Internet based on tages of SaaS are the software licensing costs are
their needs. Moreover, its architecture consists of reduced. The SaaS provider is the responsible of the
three layers that illustrate the key parts of Cloud IT infrastructure management that reduce the cost of
Computing as shown in Fig. 2. The frst layer is the software maintenance hardware fees. Also, as
appli cation layer, which includes data, where users advantages the SaaS, provide a quick set up and
can utilize diferent applications, such as web deployment. SaaS services is already installed and
interfaces. The second layer is the virtualization layer, confgured in the cloud. This reduce delays resulting
which provides the needed from often classical software installation and
confguration. Another beneft of SaaS is the
Scalability propriety, SaaS provide many subscription
confguration and fexibility to change the service
parameters when needed, for example, when you
need to add more users or more business services.
In PaaS, a platform is provided to create services
and applications with all the required resources so
that the users can create their applications.
The advantages of PaaS are data backup and
recovery and increasing the fexibility for the
developers to build new platforms to achieve newer
 requirements. In IaaS, the system administrators and eavesdropping, theft, and fraud. Therefore, the Cloud
developers can access the infrastructure of the Cloud infra structure should be secured by the providers.
service provider. They can build applications on the Monitoring the data is required because it is
infrastructure using the virtual hardware with devel vulnerable to be attacked. The data and applications
opment tools. The advantages of IaaS are reducing of the users should be protected using authentication
infra structure costs and providing on-demand methods (Qu et al. 2017).
scalability that adds more fexibility to change the Mobile cloud computing (MCC) is a combination of
organization’s require ments (Qu et al. 2017). mobile devices, mobile networks and cloud
Data security and privacy are considered as one of computing. It refers to a model whereas both data
the security risks that Cloud Computing can storage and data process ing occur outside the mobile
encounter. Criti cal data is transferred to third-party device. Mobile cloud comput ing main goal is to make
Cloud servers, which could cause security issues. applications remotely accessible by mobile devices
The security issues might be encountered by Cloud users. Example applications of mobile cloud
providers that provide (SaaS, PaaS, and IaaS) or computing are YouTube, LinkedIn, and Google+
their users. These security issues may include (Perez and
Fig. 2 Architecture of Cloud Computing (Qu et al. 2017) Kumar 2017). Through Mobile Cloud computing, mobile

13
F. Alhaidari et al.

users provided with services available on a cloud in virus scanning, and authenti
the mobile device. These services operate on mobile cation (Qi and Gani 2012; Dinh et al. 2013).
devices (smartphones and their variants) for the 3) Storage capacity/accessibility
users as if they were operating on conventional Storage is one of the main concerns in mobile
devices such as personal comput devices, people face the full-storage problem in
ers or laptops (Perez and Kumar 2017). It is predicted their mobile devices. With MCC, people can use
that by 2025, each human being will own an average sharing on cloud feature, it enables them to
of six mobile devices summing up to 50 billion more upload data such as images on the clouds and
devices (Bhullar et al. 2014). can easily access any time. It also provides
MCC has supported mobile devices with serval backup and recovery (Qu et al. 2017). There are
features which enhanced its performance to its users many examples of applications that support
and provides convenience in many ways. In the sharing on the cloud, the frst example is Amazon
following points, we dis cuss few advantages of MCC Simple Ser vice Storage, which allows users to
in general. upload fles. Another example is image photo
sharing, it allows users to upload photos.
1) Battery lifetime Examples of applications based on image photo
Enhancing CPU performance to reduce power con sharing are Flicker, Facebook (Dinh et al. 2013),
sumption require changing in the structure of and ShoZu (Oludele and Oluwabukola 2016),
mobile devices or need an advanced or new respectively.
hardware, which will result in increasing the cost 4) Dynamic provisions
of these devices (Jin et al. 2019). To solve the Dynamic provisions are an efective way for service
problem of large power com putation by avoiding providers, they can run the applications without
long application execution time on mobile resource
devices. There is a technique called consump tion reservations (Dinh et al. 2013; Oludele and
ofoading (Wu 2018), the target of this technique is Oluwabu kola 2016).
migrating large computation with complex 5) Scalability
processing from limited resources like mobile The Cloud can scale services based on the users’
devices to capable machines like a cloud (Qi and needs (Qu et al. 2017). Internet service providers
Gani 2012; Dinh et al. 2013). (ISPs) can efectively expand application and
2) Reliability service without or with a small constraint on the
With storage capability in cloud computing, people resource usage (Oludele and Oluwabukola 2016;
can easily store and retrieve data and run Shamim et al. 2015).
applications on the clouds, this feature will Few essential characteristics of MCC. These charac
eliminate the risk of data lost on mobile devices. teristics are namely the Flexibility or Elasticity,
MCC can be used as a comprehen Scalabil ity of Infrastructure, Broad Network
sive data security model for users and service Access, Location Independence, and reliability
providers. There are numbers of security services (Sharma and Soni 2013; Rajak and Shukla 2018).
can be provided remotely to the users, examples Figure 3 shows the overall architecture of a typical
of these services are: malicious code detection, mobile cloud computing environment.
 2.1 Applications of mobile cloud computing application is represented by mobile brokerage.
Mobile brokerage is a stock market service available
There are several applications of MCC (Yang and on mobile devices which enables users to interact
Hwang 2013; Alnanih et al. 2013; Gavalas et al. with the market online in a timely manner, regardless
2014; Elazhary et al. 2017a, b; Ozinga et al. 2017; of location limitations (Oludele and Oluwabukola
Harari et al. 2017), here few of them are discussed. 2016; Rountree and Castrillo 2014; Kaur and Kaur
2014; Vinh 2018). 2) Mobile healthcare
1) M-commerce Mobile healthcare ofers convenient help to access
Mobile commerce, also as known as the next gen healthcare resources such as health records
eration E-commerce, it refers to the trading of goods efciently and easily (Dash et al. 2019; Rahman
or services through internet-enabled mobile devices et al. 2018a, b; Mosa et al. 2012). Mobile cloud
such as smartphones, tablets, and PDAs. Mobile computing applications in healthcare schemes
banking is one example of mobile commerce, many are:
banks nowa days make use of mobile cloud • Enabling the healthcare providers to monitor the
computing technology to make their services easily patient at anytime and anywhere using broadband
and reachable by their clients. Through your bank wireless communication technologies. This may
mobile application, you can make online fnancial be one way or two-way communication, that is
transactions, access your account infor mation, just alert or alert and response.
purchase and stocks. Similarly, another M-com merce • Managing and coordinating emergency vehicles
when ever a call received for accident or incident.

13
Cloud of Things: architecture, applications and challenges

Fig. 3 MCC architecture (Perez rate (Rahman et al. 2018; Afolabi 2014; Rimale
and Kumar 2017) et al. 2016; Doswell et al. 2013; Lee and Kim
2013; Agca and Özdemir 2013; Karkar et al.
2017). Mobile Learning can provide many
advantages for students and teachers,
summarized as fol lows:
• Easy communication between students and
teachers. • Easy evaluation for student’s
understanding. • Timeliness and remote access.

3 Internet of Things (IoT)

Internet of Things (IoT), also referred to as the

Internet of objects, is a widely distributed network of
things. Things can be any form of different physical
objects that have unique identifers (UIDs). Also,
those objects can commu
nicate with each other without any direct human
interac tion. Since the objects or devices can transmit,
share, and use data from the physical environment,
IoT provides many

• Providing devices that sense blood pressure, pulse

rate, and alcohol level to alert the health care
emergency sys tem.
• Providing the patient or the healthcare provider with
access to current and past medical information. •
Managing healthcare costs payment and other health
related costs automatically.
3) Mobile learning
Mobile cloud computing is applied in traditional
mobile learning applications in order to solve its
prob lems related to limited educational resources,
devices’ high costs, and low network transmission
 services to multiple sectors, such as education, safety
and health. As an example, with the help of IoT,
remote moni toring of patients can be improved by
enhancing the quality of services and increased the
number of patients served at a time. In addition, IoT
is known for their use in our daily lives. For example,
sensors are one of the applications of IoT that can be
used in daily life; they can be placed in the home for
safety and better housekeeping (Sprute et al. 2019;
Sanabria et al. 2019; Ujagar et al. 2019). Appliances
control is another aspect of IoT. Along with these
evident applica
tions of IoT, security is the area concerned with
securing connected devices and networks in the
Internet of Things. It is one of the hottest areas of
research currently.
Since IoT involves the connectivity of objects to the
Internet, this allows the devices to be vulnerable to
security risks. When the IoT started to evolve and the
number of unsecured devices connected to the
Internet is increasing, security experts warned of
potential threats. For instance, when owners fail to
change the password or if changed, fail to select a
strong password of their newly purchased smart
devices might cause security risks. Similarly, in case
of hacking, the hacker may get control of the devices
that may lead to disaster. For example, in case of
trafc signals and other metropolitan applications of
IoT. Generally, there are many security challenges
that should be considered when deploying such
technology (Gloukhovtsev 2018). Examples of such
challenges are as follows:

• Scalability: Since in the IoT environment there is a

huge number of connected devices so, it needs
security solu tions that are scalable.

13
F. Alhaidari et al.

• End to end security: In order to prohibit unauthorized Several attacks against IoT have been discussed
entities from intercepting and accessing the data, such as malicious node insertion, modifcation of data
secure communication between IoT devices is and codes, availability attacks, and unauthorized
required. access. Malicious node insertion (Deogirikar and
• Connectivity: IoT needs to connect various devices Vidhate 2017) threatens the availability of IoT network
and each of these devices has diferent capabilities where the attacker exploits the hidden station
so, con necting all of these devices with diferent problem to disrupt any legitimate node. Code
capabilities must be taken into consideration. exploitation (Shrivastava et al. 2017) modifes the
• Trust and authentication: To ensure that the data is program code to make it malfunction. (Tanaka et al.
pro vided by legitimate devices and being protected 2016) discussed JTAG exploitation attack which is
against several attacks such as malicious node considered a physical attack that targets the JTAG
injection attacks. port to do several exploitations targeting IoT
• Management of identity: With weak security mecha Confdentiality, Integrity, and Availability. IoT devices
nisms implementation, an identity management can be compromised by vulnerabilities related to
issues will appear. energy consumption, memory usage, and number of
• Attack resistance: Since the IoT devices have limited IoT devices (Sandoval et al. 2019; Vishwakarma and
resources, various types of attacks that targeting Lee 2018).
the resources consumption can occur and thus it Several studies worked on addressing the security
requires strong attack-resistant solutions. aspects and challenges of IoT. To improve the
security of IoT, IoT devices should be segmented into system need to delete the legacy services or disable
its network and have net work access restricted the services that are not going to be used.
(Roman et al. 2018). In (Gope and Sikdar 2019) the (Tanaka et al 2016) discussed the availability
authors proposed a novel two factor authen tication issues in IoT such as: (1) the increasing number of
scheme for privacy-preserving of IoT devices, the IoT devices which led to increasing the attack
proposed solution allows an IoT device to perform surface. (2) Detecting attacks is harder since the IoT
anony mous communicate with the server located at system requires minimum human con
the data and control unit. The advantage of the tribution and lack of administrator. (3) The attack can
proposed scheme is it remains secure under a stay for a long time because of the long-life cycles.
physical attack with the access to an IoT device. Here, they proposed using Hitachi’s model to address
(Grifn 2017) discussed a biometrics-based such concerns. Hitachi’s model is a layered
authentica tion which is one of the techniques that architecture that has three layers including cloud,
used to enhance the security in IoT devices by sensors, and relay devices which can be a router,
enforcing the access control mechanism. The switch or gateway that collect the sensors data. The
combination of biometrics-based authen tication factor idea of the model is to detect the behaviors that are
with other factors can make the multi-factor and two normal at a specifc time and then become suspicious.
sides authentication easier and reduce the need for a They used diferent monitoring and detection models
high cost. including internal monitoring that can spot suspicious
In (Gope et al. 2019), authors proposed a behavior by monitoring the removal or insertion
lightweight mutual user authentication protocol that actions of USB and the installation of programs, trafc
authenticates the communication between a user, monitoring to detect anomaly trafc like backdoor
who is asking for accessing sessions, and cyber kill chain to evaluate the risk by
IWSN through the user trusted device, and a required utilizing the fshy action taken from the other
sensor node that is accessed through the IWSN monitoring and detecting models.
gateway. To perform such mutual authentication, they (AlDoaies and Almagwashi 2018) discussed using
proposed using lightweight cryptographic primitives Block Chain technology (BC) to secure the dynamic
including hashing and physically unclonable function IoT environ ment. BC is a distributed ledger that
(PUF) as the main components of chal contains a set of veri fed blocks that will match the
lenge and response protocol between the user device IoT’s dynamic nature and
and the gateway as well as between the gateway and exceed the traditional security frameworks’ central
the sen sor node. Moreover, they considered physical author ity problem. According to the research, the BC
layer security of the sensor nodes as a phase of the enhances the application layer’s security level, which
proposed approach where they used PUF function of can be used as a system defense versus any attack,
the sensor node as a part of authentication process minimizing the delay and overriding a single point of
and thus being able to detect any physical tempering failure issue. Moreover, since BC depends on a
on the senor nodes. digital signature and hash-function cryptographic
(Choi et al. 2018) proposed a hardening scheme to approaches, all transmitted data are digitally signed
enhance the security and minimize the vulnerability in and time-stamped to increase the reliability in the IoT
IoT devices. System hardening is a technique used to environment.
set a vari ety of functions in a system to enhance (Yasir et al. 2016) discussed the use of Social IoT
security and reduce the threats and security (SIoT) recommendation services between diferent
vulnerabilities, for example, OS hardening in the IoT applications.

13
Cloud of Things: architecture, applications and challenges

SIoT uses a perception layer to sensing and gathering concerns in detail includ ing integrity, confdentiality,
infor mation from IoT devices and then IoT devices availability as well as scalabil ity issues. These
can create social relationships among themselves concerns are resulted from the capacity and
based on the collected data. In order to use this data, computation power limitations of IoT nodes which
information is forwarded to the network layer and make deploying regular security protocols impractical
then forwarded to the upper layer for storage. Finally, on IoT environment. However, several studies were
data can be used by the social IoT recom mendation discussed in this paper on addressing some security
system to build and maintain social relations and concerns by propos ing diferent security techniques
profles. considering light-weight protocols that can be applied
As a summary, IoT provides many applications and on IoT environment. Among these techniques
car rying their activities through internet connections discussed in this paper are: (1) techniques for
and thus one of the main challenges and concerns is authentication such as two factor authentication
the security risks. Many studies have discussed such schemes (Gope and Sikdar 2019), a biometrics-
based authentication and access control (Grifn 2017), process between application and data. In IoT,
and lightweight crypto graphic mutual user collect ing data and sharing them can be ofered at
authentication (Gope et al. 2019). (2) Techniques that low-priced through automation. Also, it ofers
rely on monitoring and detection methods such as custom-made uni versal application. On the other
hardening the OS systems (Choi et al. 2018) and hand, Cloud employs personalized and integrated
Hitachi’s monitoring model for detecting abnormal apps by employing these apps, the Cloud can
behav iors within IoT environment (Tanaka et al. reach and run whatever and when ever they want.
2016). (3) Tech niques based on application layer’s Likewise, the outcome for the connec tion in Cloud
security level such as BlockChain technology for is adequate and inexpensive. It is true that
providing more reliability in the IoT environment integrating Cloud to IoT will enhance the
(AlDoaies and Almagwashi 2018) and trusting model connectivity/ accessibility of devices, but it may
based on Social IoT (SIoT) recommendation services cause some issues considering specifc
(Yasir et al. 2016). circumstances (Aazam et al. 2014; He et al. 2014;
Wang and Deters 2009).
2) Storage
4 Cloud of things IoT contains a massive amount of information source.
It performs collecting, processing, and sharing a
Day by day, the amount of data is expanding, soon large amount of data. Cloud is the most efective
will be a must to use external storage to store the solution to provide an on-demand unlimited virtual
interacted devices data. Furthermore, knowledge can low-cost storage to the data produced by IoT
be extracted from the stored data to provide valuable (Aazam et al. 2014).
information. Sadly, this trig 3) Computation
gers the need for additional processing on the IoT does not allow complex locality processing, which
external stor age, which Cloud Computing provides is a limited processing feature. Whereas Cloud
along with unlimited computational capability. Comput ing provides unlimited processing (Aazam
However, this feature is applied in IoT in a limited et al. 2014). B. Architecture design for CoT
manner and IoT devices can take advan tage of the CoT architecture allows dynamic addition of IoT
unlimited storage and computational capabilities devices as depicted in Fig. 4.
available on the cloud servers. IoT deals with real- The idea of such architecture is to provide a frame
world things to the contrary Cloud Computing cope work facilitating the integration of various
with virtual resources. Cloud can take advantage of components and layers to utilize the power of
the feature that IoT covers and deals with the real- cloud computing for processing, managing, and
world thing, reaching realistic scenarios. This brings transforming information related to IoT
us to the integration of both Cloud Computing and environment. Figure 4 shows the key lay ers,
IoT, now called “CoT” which stands for namely sensing layer, communication layer,
“Cloud of Things.” (Stergiou et al. 2018; Aazam et al. control layer, and actuation layer.
2014). It is also termed as CloudIoT in some The frst layer, sensing layer, is responsible of gath
researches (Abdelwa hab et al. 2016a, b; ering information from various sensors that can be cat
Krishnamurthy 2016). egorized based on the sensor type (homogeneous or
a heterogeneous), sensor methodology (like active or
A. Drivers of the integration passive), or number of sensing parameters (one
Since both IoT and Cloud Computing can complete param eter like body temperature, many parameters
each other, there are drivers lead to generate such like sig nals generated by electrocardiogram (ECG)
inte gration, that comes in three categories system). This layer involves sensor and recognition
(Aazam et al. 2014). technologies that provides information about people,
1) Communication objects, and environmental conditions. It represents
The key feature of Cloud and IoT are the sharing sources of the

13
F. Alhaidari et al.
Fig. 4 CoT architecture design (Abdelwahab et al. 2016a, b) considering the human

information that will be communicated, processed,

and then delivered in a form that can be utilized to interaction, or automatically. Thus, for manual
provide knowledgeable and meaningful actions in case, this layer should represent the information
the physical world. in human read able format whereas in case of
The second layer, communication layer, represents automatic procedure it should be represented in
the core and access network layer responsible of form of commands that ft desired actions in the
main taining communications among various physical environment.
elements of the IoT including sensors, things and
humans. This involves diferent networking
approaches and technologies which could be as
an example wireless sensor networks (WSN), 5 Cloud of Things issues
body sensor network (BSN), or net generation and possible solutions
networks (NSG). The network layer comprises of
gateways that receive and combine the It will not be that simple to have all the resources
unstructured data coming from the sensing layer available over Cloud Computing and let everything to
and then transmit it to the control layer where it become part of IoT. The integration of these two
can be stored and processed by the cloud com technologies generates some serious issues that
puting resources. need to be considered. Some of these issues will be
The third layer is the control layer which is responsi discussed below (Stergiou et al. 2018). Moreover,
ble of storing and processing the transmitted data as Table 1 contains the list of these issues and their
well as it provides various services for the IoT brief description.
components such as hosting, monitoring, and
managing devices. 1) Energy efciency
Finally, the fourth layer in this architecture is the actu Energy efciency will be afected by the connectivity
ation layer. This layer is responsible of representing with the Cloud and the sensor networks, which
the information received from the control layer by causes heavy data communication that needs a
making it in a form that can be used to perform lot of power. Therefore, there should be
actions as a response to sensing layer inputs. The environmental power sources that provide a
actuation proce dure can be performed manually permanent power supply (Stergiou et al. 2018).

13
Cloud of Things: architecture, applications and challenges
Table 1 CoT issues
Issue Brief description

Energy efciency Data communication require a lot of power

Resource allocation Allocation of appropriate resources for CoT is difcult
Identity management Allocation of unique ID for CoT objects and mobile devices is difcult Service discovery
Discovering and updating service is difcult
Location of data storage Multimedia data cannot be stored in virtual storage
Security and privacy Lack of trust in service providers and the lack of knowledge about the data physical location Communication
of unnecessary data Consuming a lot of power, storage space, and processing Heterogeneity Variety of devices with their diferent
operating systems, platforms and services Performance The aspects related to communication, computation and storage Big data
Big data communication, storage, access mechanism and processing

Solution: Following types of platforms ofer low devices.

power and low-cost connectivity management • Determine an identity lifecycle for CoT devices,
solutions either by Wi-Fi or cellular technologies. making sure it can be modifed to meet the
This can vary from hardware connectivity, cellular projected lifetime of these devices.
networks, to the data routing features of the same • Create a well-defned process for registering the IoT
set of sensor data (Bak devices; the type of data that the device will be
tayan and Zahary 2018). Examples: Mulesoft transmit ting and receiving should shape the
(MuleSoft 2019), Hologram (Hologram 2019), registration process.
Sigfox (SigFox 2019). • Defne security safeguards for data streams from IoT
2) Resource allocation devices.
It is difcult to estimate how much does an entity or a • Outline well-defned authentication and authorization
particular IoT would require a resource on the process for admin local access to connected devices.
Cloud. Mapping resource allocation depends on • Create safeguards for protecting diferent types of
the sensor and its purpose, type, amount, and data,
frequency of data genera making sure to create privacy safeguards for
tion (Stergiou et al. 2018). personally identifable information (PII).
Solution: Sensor node should transfer data packet • Defne and document who can access the diferent
along with its frequency. Moreover, Fog computing types of data (e.g., local admins, outside companies).
is preferred with devices that demand a lot of • Establish guidelines for authenticating and
resources. In such devices, the procedures authorizing occasionally connected CoT devices.
cannot be uploaded to the cloud. Hence, this • Determine procedures for access control to IoT
results in dropping devices’ complex devices and system based on the company’s
ity, lifecycle costs, and power consumption policies.
(Baktayan and Zahary 2018). 4) Service discovery
3) Identity management With the Cloud of Things, the responsibility for dis
A unique identifcation is needed for objects when covering new services for the users rely on the
they are becoming part of the Internet (IoT). Also, manager. However, any object can be a part of IoT at
when Mobile devices entered a new network, they any moment and leave at any moment. So,
need to have identity mapping in the network. discovering new services, their status, and update the
Since even this kind of ubiquitous networking can service advertisement would be an issue. To solve
be supported by IPv6 address, assigning IPv6 this issue, a uniform service dis covery way would be
addresses can be an acceptable way in this required (Stergiou et al. 2018).
regard (Qi and Gani 2012). Solution: Uniform service discovery is preferred. Tra
Solution: Setting IPv6 address to all of them. The ditional web service discovery approaches and their
growing world of Internet of Thing has impacted used technologies are unsuitable for the discovery of
how identity and access management (IAM) need CoT ser vices. This is due to the diference between
to work. In the Identity of Things (IDoT), IAM must CoT services and traditional web services. The
be able to man services in the IoT are linked directly to the physical
age human-to-device, device-to-device, and/or world because they are provided by objects in the
device to-service/system. CoT implementation real world, they are deployed in resource constrained
strategies must include a method for managing devices with limit capabilities also they are operated
device identity and be in highly dynamic environment. All these features
able to do some of the following (Dim make the task of locating the desirable services
2019): • Establish a naming system for CoT difcult to be accomplished. The existing ser
 13
F. Alhaidari et al.

vice discovery approaches do not efectively adapt damaged and leaked. Moreover, CoT security
to the features of CoT services, especially that considerations must
there is no new standard language for describing contemplate security from multiple facets (Albishi
the CoT services in a formal way, and concerning et al. 2017). For example:
the solution of extending the traditional web Data at rest: Data housed in applications and data
services technologies like OWL-S with new bases on-premises or in the Cloud is said to be “at
attributes or WSDL with DPWS or COAP services rest.” Most organizations rely on conventional
has to take in consideration the balance between perimeter-based defenses, such as frewalls and
to be lightweight to support the limit capabilities of anti-virus programs, to protect data at rest.
devices and to adapt to the IoT services features However, hackers fnd these troves of data
(Aziez et al. 2017). irresistible; hence, the Broadband Internet Techni
5) Location of data storage cal Advisory Group and Cloud Security Alliance
For critical, latency, and time-sensitive data, location (BIT AGCS) recommend employing a combination
is very important. Therefore, it is essential to store of hard ware and software encryption techniques
the multimedia data in a physical location that is to ensure the security and integrity of data at rest.
close to the user to minimize the time needed to Data in use: Data “in use” by an application or gate
access them, which will lead to minimizing the way must be accessible to users and devices,
latency (Stergiou et al. 2018). making it the hardest form of data to secure. With
Solution: Time-sensitive data should be allocated. in-use data, security depends on the strength of
Moreover, as a solution, Infrastructure-as-a- authentication pro cedures and the number of
Service (IaaS) providing IoT-compatible technical users and devices accessing the data.
loads and backend services for example, extract, Data in fight: The data in the channel, such as from
transform and load (ETL) processes which the device to the Cloud. In this regard, well-
synchronize to connected devices and data; established Internet communication protocols
Example: Azure (Microsoft Azure 2019), Amazon armed with modern cryptography algorithms make
Web Services (AWS) (Amazon 2019) and it virtually impossible for hackers to decipher data
Salesforce (SalesForece 2019) etc. in transmission. While many IoT devices support
6) Unnecessary communication of data multiple security protocols, few enable them as
At some point when the data is not required, upload part of their initial confguration. At a minimum, IoT
ing them to the Cloud will not be necessary. In this devices that connect to mobile applica
situ ation, gateway device must determine when it tions or remote gateways should employ HTTPS,
is needed to stop the data uploading or the data trans port layer security (TLS), secure fle transfer
generation, so the resources of the Cloud and protocol (SFTP), DNS security extensions, and
network will not be con sumed, that will also other encryption protocols. Moreover, separating
optimize the utilization of power. For this purpose, information-only data from action data—using
the gateway device should have the ability to do encrypted, one-way, outbound communications—
extra processing before sending it to the Cloud limits vulnerability should the data be intercepted
(Stergiou et al. 2018). while in fight.
7) Security and privacy 8) Heterogeneity
As mentioned before in this paper, security is a major Existence of several devices having diferent oper
concern in both IoT side and Cloud side. The ating systems, vendors, following diferent
integra tion between them expected to operating standards is another possible issue in
compromise security and to be exposed to CoT (Sahmim and Gharsellaoui 2017). To
diferent security attacks, such as session riding, mitigate this issue, either the devices can be
SQL injection, cross-site scripting, and side- selected from same vendor which is not possible
channel, in addition to several vulnerabilities, such in many cases. Otherwise selection of appro priate
as session hijacking and virtual machine escape. middleware and protocols can help reducing this
Simi larly, privacy would also be an issue; issue which may end up with poor performance.
concerns will arise when applications move into Solution: Appropriate protocols and middleware can
the Cloud due to the lack of trust in service be in place. Moreover, fully managed IoT
providers and the lack of knowledge about the deployments including standard hardware,
physical data location. So, critical or sensitive confgurable software, connectivity services, and
information should be stored virtual storage server device management for exam
that is in a known and trusted physical location ple, over-the-air (OTA) upgrades, for example:
(Stergiou et al. 2018). Particle (Particle 2019). In addition, Fog nodes
Solution: Critical data prevented from being are designed by diverse constructors’ providers
 and come in various pat terns and demand to be diferent platforms.
deployed depending on their relevant platforms. 9) Performance
So, the fog nodes have the capability to work on

13
Cloud of Things: architecture, applications and challenges

The performance aspect of CoT depends upon vari 10) Big Data
ous factors like communication links being used Whenever, it is thought of cloud computing, Big data
for data transfer, computational efciency related is an essential part of provided services.
services of associated cloud and storage access- Moreover, the data being collected from IoT
delay etc. To overcome this issue, fog, edge and devices can be integrated and analyzed by the
mist computing breakpoints can be utilized so that cloud. The possible issues associ
the workload can be balanced over certain links ated with Big data that can contribute to CoT as
and overall performance can be improved. well are mainly, data communication, storage,
Solution: Edge, Fog and Mist computing may be access mecha nism and processing of data. The
investigated. In addition, the practice of using a countermeasures in this regard are somewhat like
network of remote servers hosted on the Internet the performance related issue.
to store, man age, and process data, rather than a Solution: Appropriate middleware processing tech
local server or a per sonal computer. Cloud niques can be used. Moreover, tools managing
Computing can be a heavyweight and dense form data rout ing from IoT devices can play a
of computing power and it is an integral part of signifcant role. In addi tion, third-party services, as
CoT. Moreover, Fog computing is classifed data well as analytics to extract
processing toward the CoT devices. The fog insights from IoT data can be a plus for example,
permits data processing to be done relying on Clear blade (ClearBlade 2019) and ThingSpeak
application needs, available networking, and (ThingSpeak 2019) etc.
computing resources. Hence, fog cuts the total of
data demanded to be sent to the cloud, which
saves network bandwidth. Summary
Only required data should be distributed and com Section 5 presented a summary of possible
municated. Moreover, solutions for the following issues: Energy Efciency,
• Identify and classify data collected throughout your Resource Alloca tion, Identity Management, Service
CoT ecosystem and learn their corresponding Discovery, Location of Data Storage, Unnecessary
business use case. Communication of Data, Secu rity and Privacy,
• Identify and execute on opportunities to stop Heterogeneity, Performance, and Big Data. As a
collecting unused data or adjusting summary to the CoT issues and their possible
their granularity and retention time. solutions, it is worth mentioning that both factors are
• Consider using tokenization and one- subject to the type of application and domain where
way cryptographic hashing wherever you don’t the CoT is deployed. It is not necessary that all the
need specifc data in its entirety. issues may arise in all type of applications/domain.
• Consider using asymmetric cryptography to protect This section just summarizes and com prehends the
data at rest on CoT devices and devices that are possible issues and possible solution a typi cal CoT
only responsible for temporarily collecting and environment may have. So accordingly, a suitable
batching data and periodically submitting the data solution may be chosen/adapted by the service
to other systems for processing. providers or the solution experts. As far as the cost of
• Only store and transmit data to central systems with deployment is concerned, depending upon the
strong ownership and strict security controls. requirements, trade-ofs can be made between
• Follow the principle of least privilege in granting various solutions.
access to any collected data.
• Identify and consider the unique capabilities of your
CoT devices. This could include mobility, actuation,
sensory data collection and transmission, and 6 Cloud of Things applications
ownership trans fers that impact your regulatory and
legal compliance. The emerge of Cloud of Things technology is making
• Consider privacy and transparency expectations of a sig nifcant impact on the way people are living their
your customers and corresponding legal every day life. A generic architectural framework of
requirements in the jurisdictions where you CoT is given in Fig. 7 (Abdelwahab et al. 2016a, b). It
manufacture, distribute, and operate your CoT is comprised of three tiers namely the Cloud tier, the
devices and systems. cloud agent tier and the CoT tier. This framework is
 equally useful for all the possible applications like in the form for sensors’ output.
healthcare, smart cities etc. men tioned afterwards. In Fig. 7, Sensor network virtualization in sensing
The major set of associated CoT devices assumed as a service using diferent cloud agents near the
are the sensing devices that are widely used in edge. Arrows and numbers illustrate the diferent
healthcare, vehicular networks and other wireless steps and messages sequence. Five main steps are
sensor networks. Consequently, the data collected is involved in terms of service acquisition.

13
F. Alhaidari et al.

• Firstly, tier clouds are conventional cloud computing vices of IoT systems are managed to generate a
platforms, and cloud agents are edge computing business model and fowcharts, which are done on
plat form with evolved rule for sensing as a service the business layer (Ghanbari et al. 2019).
the cloud user will submit the request duly Below are some application benefting from Cloud of
received by the frst-tier clouds. Things are listed along with a short description and
• Secondly, the frst-tier clouds send the request to the potential challenges related to each application
appropriate associated cloud agent. This optimized (Aazam et al. 2014; Uma et al. 2017).
device virtualization will serve as virtual sensor
networks by exploiting the discovered sensing 1. Healthcare
resources In health sector, Cloud of Things is used to simplify
• At the third step, the cloud agent communicates with the healthcare process and the quality of the
the cloud of thing network associated with service given. Through diferent wearables, such
corresponding set of CoT devices. as watches, and other sensors data is collected
• At the fourth step, this information is sent back to frst on a regular basis, and the integration of the
tier clouds via the same agent referred to in the step Cloud enables having all the heavy algorithms run
3. • Finally, the frst-tier cloud submits the response to on the Cloud especially that these wearables
the user after due processing on the data delivered have a limited computational capacity and battery
by the CoT tier. (Aazam et al. 2014). Reports then are sent to the
authorized healthcare providers and patients to
The frst tier plays a role of receiving the request from help notice any changes in the patients’ health.
Cloud users as a sensing task and then reformat it to This beneft is noticeable especially in ambient
be processed and interpreted later by Cloud agents. assisted living systems that are aimed at people
The sens ing task should hold parameters that allow with chronic medical conditions and disabilities
Cloud agents to decide about the source of required (Carr and Doleac 2016). Figure 7 shows a typical
information, query parameters, CoT category, and CoT based healthcare architecture (Dang et al.
other parameters that might support and clarify user 2019). It is mainly comprised of three layers, akin
task requirements. To do such pro cess, Cloud tire to the concept given in Fig. 5. The sensors, the
involves integration of several resources including agents and the processing clouds. The sensors
storages, analysis, and transformation tools. Based are the CoT devices collecting sensory data like
on the sensing task, Cloud agent captures data from body sensors, patient health record
CoT tier and return results back to the Cloud tier (PHR)/electronic medical record (EMR) data
which in turn can process, analyze, and store such (Mahmoud et al. 2018; Bharadwaj et al. 2017;
data to send required results back to the Cloud user. Medina et al. 2018; Baig et al. 2015) and remote
The underlying CoT frame work includes the physical monitoring data
sensors and actuators, which are referred to the
perception layer (CoT). The produced data in this
layer is passed to the middleware layer (cloud agents)
by a network layer. The secure channels for moving
data use

Fig. 5 A generic CoT frame

work for SaaS (Abdelwahab
et al. 2016a, b)
some technologies such as RFID, Wireless Sensor
Network (WSN), Wi-Fi, Worldwide Interoperability for
Microwave Access (WiMAX), and Long-Term
Evolution (LTE) with its requester. The application
layer (frst tier clouds) provides the requested services
for customers. The activities and ser
13
Cloud of Things: architecture, applications and challenges

etc. This is duly processed at the associated of that would not be possible without the help of
clouds and conveyed by the agents. both technologies (Aazam et al. 2014; Korczak
Challenges: Privacy, legal and social aspects, large and Kijewska 2019; Iwan and Malecki 2016;
scale, security, reliability, performance, and heteroge Jabeur et al. 2017; Karakikes and Nathanail 2017;
neity (Aazam et al. 2014; Munir 2018; Chowdary and Witkowski 2017). Figure 6 shows a typical
Yakobu 2016; Farahzadi et al. 2018; Patel and architecture of smart homes (Razvi et al. 2015).
Parmar 2016). To overcome this issue, advanced As mentioned earlier in Fig. 7, the three tiers are
security stand ards should be followed for end to end appar ent. Namely, the CoT based devices to
encryption. 2. Smart homes sense the data, the agents to convey the data and
In smart homes, many technologies beneft from IoT the cloud layer to process the data.
and Cloud Computing. Lightning, heating, Challenges: Reliability, performance, and heteroge
controlling the windows blinders and making a neity (Aazam et al. 2014).
person’s morning cofee (Cubo et al. 2014). In all 3. Smart logistics
these applications, user’s data is collected and
analyzed and is communicated with all these
devices to make the ultimate experience a user

Fig. 6 Architecture of smart

homes

Fig. 7 A CoT based healthcare

framework
can have. It also saves energy and helps the
environ ment by turning of unused devices, and all
 13
F. Alhaidari et al.

It enables tracking goods and shipments between logistics

source to destination using geo-tagging just two examples, and there are many more CoT
technologies (Aazam et al. 2014). Figure 8 shows appli cations in developing smart cities.
a typical architec ture of smart logistics using CoT. Challenges: Privacy, legal and social aspects, secu
The devices originate the data from the rity, reliability, performance, and heterogeneity
associated sensors, passed to the cloud for (Aazam et al. 2014).
processing via gateway (Pattama 2018). As far as 5. Video surveillance
the analysis part is concerned, it might be part of Advance video surveillance in this era performs high
cloud or some middleware. computational algorithms and complex video
Challenges: Legal and social aspects, large-scale, analysis. Home video surveillance allows its users
reliability, and heterogeneity (Aazam et al. 2014). 4. to stream those videos from their mobile devices.
Smart cities All of this is made possible through Cloud
The Cloud of Things opens the world to create new Computing (Atlam et al. 2017). However, this
opportunities for contextualization and geo- change will be slow because of the new hardware
awareness. Taking the data collected from that has to be installed replacing millions of old
sensors around the city region and transform it to cameras.
intelligent systems to help make better decisions Challenges: Security, reliability, performance, and
on diferent aspects to facilitate and further equip heterogeneity (Aazam et al. 2014).
the city. ShotSpotter, for example, that alerts 6. Automotive and smart mobility
police or relevant authorities in case of fr The Cloud of Things ofers the opportunity to address
ing along with the exact location of incident. It heavy trafc, congestion, and vehicle safety. Two
greatly helped in timely report of crimes especially ways of communication can occur vehicle to
in areas where crimes are mostly underreported vehicle, and vehicle to infrastructure. All these
(Uma et al. 2017). The working model for communications result in an abundant amount of
ShotSpotter is given in Fig. 9. Smart Streetlights information that gets uploaded to the Cloud for
that are packed with sensors that detect trafc fast and accurate responses and analysis (Jin
congestion and dangerous intersections and et al. 2016).
weird noises that might indicate that a crime or a Challenges: Security, reliability, performance, and
problem is happening. It also helped in saving the heterogeneity (Aazam et al. 2014).
energy costs by automatic switching on and of the 7. Environmental monitoring
streetlights based on some activity (Atlam et al. The Cloud of Things technologies help in monitoring
2017). These are and protecting the environment. Sensors and
actuators

Fig. 8 Architecture of smart

13
Cloud of Things: architecture, applications and challenges

Fig. 9 Working model of Shot mance, and heterogeneity (Aazam et al. 2014).
Spotter (Uma et al. 2017)

6.1 Challenges in Cloud of Things

As been said, there are benefts and reasons that lead

to the integration namely the CoT. However, there
are certain chal lenges and concerns regarding this
integration, the consist ency, and accessibility of the
CoT service provided along with the security of the
data are concerned to be one of CoT critical
challenge (Botta et al. 2016).

1) Security and privacy

In the cloud environment, there is the concept of iden
tity management (IdM) which is the performing of the
methodology of creating, managing and using the
user identities and infrastructure which provide
support to the set of these processes. IdM systems
consist of users that want to access resources,
identities which are called Personally Identifable
are placed in the same area; sensors collect the Information (PII) and used to identify the attributes of
data, send it to the Cloud that does further a user, identity providers (IdP) that carry out the
processing and analysis, then it asks the authentication process and it is also
actuators to perform specifc tasks. These
technologies are used in the monitoring of water
level, gas concentration in air, soil humidity and
other characteristics, infrared radiation for fre or
animal detection. The Cloud aspect also ofers
faster response between end user and sensors
(Aazam et al. 2014).
Challenges: Large-scale, security, reliability, perfor
 the one which manages and distribute user PIIs.
Lastly is the Service Provider or SP which provide
services. The use of IdMs is a practical solution
for many problems, however integrating the IdM
with the cloud still a new feld which needs further
searching.
Another issue regarding cloud computing is privacy
where cryptography can help Cloud computing to
be safe and secure storage. Symmetric-key
algorithms, Asymmetric-key algorithms, and
Hashing are the three collections that makeup
cryptography which will be used in this approach.
Encryption will aid the users in preventing such
abuses/attacks such as man-in-the
middle, spoofed attacks, and session hijacking that
will be treating privacy. These crypto algorithms
will be designed so it ensures usability and
security. Hence, Cloud data management cannot
be fully trusted by data owners. The data process
and computation could expose the privacy of
users. That is the reason that amplifes why we
need cryptography in the frst place for over
coming the above problems.
The Federated Cloud is when the service provider
uses the individual’s cloud providers that are work
seamlessly and collaboratively in order to form a
fed eration cloud. It is considered a federation of
identities, assets, and confguration of multiple
clouds in a singles resource to make the cloud
computing practical. The benefts of using the
Federated Cloud are minimizing the risk of data
interference, confrming the data integrity, and
reducing the loss of service availability. The Feder
ated Cloud Identity Management (FCIDM) system
is a solution which manages the identities in the
cloud by

13
F. Alhaidari et al.

allowing the right individuals at the right times to investigated. Like encryption standards, intrusion
access the right resources for the right reasons. detection, frewalls and other state of the art
Moreover, there is an unavoidable threat has been surveillance mechanisms. Since security and
considered under some two reasons, one of them privacy is a big area to cover in CoT, the
is that there is a massive number of connected techniques and methodologies are also great in
devices in CoT. Second, is the key management, number (Sahmim and Gharsellaoui 2017;
public key infrastruc DeDonno et al. 2019). Following is the gist of
ture is unable either to manage the load or either such approaches other than those mention.
the storage of it. As been knowing that CoT has a • Steganography
lot of connected objects, because of that applying • Digital watermarking
public key cryptosystem neither do the old- • Obfuscation
fashioned cryptography is not achievable in CoT • Anonymization
even when applying it with strict calculation. In • Sticky policy
conclusion, PKI cannot be applied regarding the • Trusted platform module
CoT (Botta et al. 2016; Renner et al. 2014). • Data segmentation
Suggested solution: Instead of using one standard • Trusted third party mediator
for end to end channel, the use of multiple • Identity and access management
technologies in diferent segments can be • Key and security management
2) Heterogeneity communication models like Bluetooth and ZigBee
Before the integration of the Cloud, IoT applications by injecting them into a device Cloud.
and devices were thought of as isolated vertical Middleware at runtime and protocols in the gateway
solu tions, in which all system components are are to be regulated by a mapping approach.
tightly cou pled to the specifc application context. Moreover, to fexibly handle huge number of IoT
However, with the introduction of the Cloud, devices over the cloud, IPv6 address space can
diferent IoT applications will have the ability to be a practical choice which supports pervasive
communicate with each other if the heterogeneity networking and can be more reasonable in the
problem is solved. This challenge is amplifed if dynamic environment.
the user chooses multiple Cloud services to 5) Performance
increase the performance of the system (Aazam There are many qualities of service measures that
et al. 2014). are requested from CoT applications, such as
Suggested solution: To overcome this issue, one communica tion and computation speed and
pos sible solution is to investigate and provide an storage consumption also the access control
automatic device integration mechanism. That will mechanism (the time it takes to get the
start building the process from one end to other credentials verifed). Among these, the most
while resolving issues at each segment by a challenging one is communication for obtaining a
middleware semantic interoper ability (Heiler sta ble, acceptable network performance to reach
1995). the Cloud (Aazam et al. 2014).
3) Reliability Suggested solution: This issue can be resolved by
In critical applications, such as HealthCare applica provision of device bufering and some modifed ver
tions, this challenge is signifcant. Although when sion of CODA for disconnection handling (Kistler and
intro ducing the Cloud, it ofoaded heavy Satyanarayanan 1992) and other efcient synchroniza
computational work tion techniques for the virtual databases may be
from the IoT to the Cloud; it created uncertainty, investi gated (Li et al. 2017; Kekgathetse and Letsholo
which is the availability of the datacenter itself 2016). 6) Legal and social aspects
(Aazam et al. 2014). When building a system based on devices that are
Suggested solution: This is mainly associated with owned by diferent users and storing it in the Cloud,
Cloud side availability. Then Cloud resource manage there are two aspects: The Cloud service provider
ment system must be upgraded to consider the must adhere to international law and users must
Quality of Service required by the IoT devices at a agree to share their data to help build and
certain time. 4) Large scale improve the system (Aazam et al. 2014).
The Cloud has enabled real-time processing of the Suggested solution: A trust agreement can be estab
data originating from the huge number of IoT lished before each user/device integrated to the
devices. Sometimes, these devices can reach CoT. It can be a modifed/updated version of
hundreds of thou sands creating storage and tradition service level agreement (SLA) used in
communication delay issues that all must be Cloud Computing envi ronments. (Govindaraj and
tackled (Aazam et al. 2014). Jaisankar 2017; Wu 2018; Horvath and Agrawal
Suggested solution: To mitigate the scaling issues, 2015; Hussain et al. 2014; Marud hadevi et al.
one possible solution is to implement 2015).

13
Cloud of Things: architecture, applications and challenges

Summary: Section 6 presented a summary for applications are also enumerated along with their
each of the discussed challenges in CoT including suggested solutions in general.
Security and Pri vacy, Heterogeneity, Reliability,
Large Scale, Performance, as well as Legal and
Social Aspects. Moreover, this sec tion encompasses
7 Conclusion
the potential applications/application areas where
CoT can be utilized efectively. In this regard, seven
From the impact that it has, the Cloud of Things is
diferent application areas are mentioned. Although
becoming the next big thing in terms of Information
there could be limitless application areas like Industry
and Communica tion Technology (ICT). In this paper,
4.0 standard is purely based on these advanced
frstly, the mobile and cloud computing (MCC)
technologies like CoT (IoT+Cloud), Artifcial
technologies and the IoT technol ogy alongside their
Intelligence, 3D printing, augmented reality,
benefts and issues are discussed. Sec ondly, the
blockchain etc. (Machado et al. 2020). Later, the
Cloud of Things (CoT) is introduced along with its
possible challenges that may be faced during these
drivers of integration, the architecture, and the international conference on internet of things (iThings) and
possible issues in integration. Finally, a potential list IEEE green computing and communications (GreenCom) and
IEEE cyber, physical and social computing (CPSCom) and
of applications is discussed where CoT seems IEEE SmartData
promising and the challenges it may face. In addition Aziez M, Benharzallah S, Bennoui H (2017) Service discovery
to that, some general challenges to the CoT are for the internet of things: comparison study of the
debated and their possible solutions are also approaches. In: Proceedings of 4th international
conference on control, decision and information
proposed. The researchers and scholars of related technologies (CoDIT), Barcelona, pp 599–604
areas are encouraged to explore these challenges, Baig MM, GholamHosseini H, Connolly MJ (2015) Mobile
investigate the pro posed as well as other possible health care applications: system design review, critical issues
solutions to help the imple menters and developers of and challenges. Australas Phys Eng Sci Med 38:23–38
the CoT applications, to mitigate them. Baktayan A, Zahary A (2018) A review on cloud and fog
computing integration for IoT: platforms perspective. EAI
Endorsed Trans Internet Things 4(14):1–17
Bharadwaj SA, Yarravarapu D, Reddy SCK, Prudhvi T,
Sandeep K, Reddy OSD (2017) Enhancing healthcare
References using m-care box (monitoring non-compliance of
medication). In: Proceed ings of the 2017, international
Aazam M, Hung PP, Huh EN (2014) Cloud of ihings: integrating conference on I-SMAC (IoT in social, mobile, analytics
internet of things with cloud computing and the issues and cloud) (I-SMAC), Palladam, India, pp 352–356
involved. In: Proceedings of 11th international bhurban Bhullar J, Mancilla A, Nijjar A, Teixeira A (2014) The future of
conference on applied sciences & technology (IBCAST) mobile computing in 2025. Storify. https://storify.com/mobil
Islamabad, Pakistan, pp 414–419 ecomputing/the-future-of-mobilecomputing-in-2025.
Abdelwahab S, Hamdaoui B, Guizani M, Znati T (2016a) Cloud Accessed 11 May 2020
of things for sensing-as-a-service: architecture, Bittencourt L, Immich R, Sakellariou R, Fonseca N, Madeira E,
algorithms, and use case. IEEE Internet Things J Curado M, Villas L, DaSilva L, Lee C, Rana O (2018) The
3(6):1099–1112 internet of things, fog and cloud continuum: integration
Abdelwahab S, Hamdaoui B, Guizani M, Znati T (2016b) Cloud and challenges. Internet Things 3(4):134–155
of things for sensing-as-a-service: architecture, Botta A, Donato W, Persico V, Pescapé A (2016) Integration of
algorithms, and use case. Proc IEEE Internet Things J cloud computing and internet of things: a survey. Future
3(6):1099–1112 Gener Comput Syst 56:684–700
Afolabi OA (2014) On mobile cloud computing in a mobile Carr JB, Doleac JL (2016) The geography, incidence, and
learning system. J Inf Eng Appl 4(5):82–87 underreport ing of gun violence: new evidence using
Agca R, Özdemir S (2013) Foreign language vocabulary shotspotter data. In: Pro ceedings of APPAM big data
learning with mobile technologies. Proc Soc Behav Sci 83:781– pre-conference workshop, pp 1–17
785 Albishi S, Soh B, Ullah A, Algarni F (2017) Challenges and Choi S-K, Yang CH, Kwak J (2018) System hardening and
solutions for applications and technologies in the internet of security monitoring for IoT devices to mitigate IoT security
things. Proc Comput Sci 124:608–614 1 vulnerabili ties and threats. In: KSII transactions on
AlDoaies BH, Almagwashi H (2018) Exploitation of the promis internet and information systems, Korea
ing technology: using blockchain to enhance the security Chowdary ED, Yakobu D (2016) Cloud of Things (CoT)
of IoT. In: Proceedings of 21st Saudi computer society integration challenges. In: Proceedings of 2016 IEEE
national computer conference (NCC), Riyadh, 2018, pp international con ference on computational intelligence
1–6. https://doi. org/10.1109/NCG.2018.8593102 and computing research (ICCIC), Chennai, pp 1–5
Alnanih R, Ormandjieva O, Radhakrishnan T (2013) Context- ClearBlade (2019) https://www.clearblade.com/. Accessed 9
based and rule-based adaptation of mobile user Dec 2019 Cubo J, Nieto A, Pimentel E (2014) A cloud-based
interfaces in mhealth. Proc Comput Sci 21:390–397 internet of things platform for ambient assisted living. Sensors
Alrawahi AS, Lee K, Lotf A (2019) A multiobjective QoS model 14(8):14070–14105 Dang LM, Piran MJ, Dongil H, Kyungbok
for trading cloud of things resources. IEEE Internet Things M, Hyeonjoon M (2019) A survey on internet of things and
J 6(6):9447–9463 cloud computing for healthcare. Electronics 8(7):768
Amazon (2019) https://aws.amazon.com/iot/. Accessed 9 Dec Dash S, Biswas S, Banerjee D, Rahman A (2019) Edge and fog
2019 Atlam H, Alenezi A, Alharthi A, Walters R, Wills G (2017) com puting in healthcare - a review. Scalable Comput
Integra tion of cloud computing with internet of things: Appl Exp 20(2):1–30
challenges and open issues. In: Proceedings of IEEE

13
F. Alhaidari et al.

DeDonno M, Giaretta A, Dragoni N, Bucchiarone A, Mazzara M DIM (2019) https://www.digicert.com/internet-of-things/device-

(2019) Cyber-storms come from clouds: security of cloud ident ity-management.htm. Accessed 9 Dec 2019
com puting in the IoT era. Future Internet 11(6):127 Dinh HT, Lee C, Niyato D, Wang P (2013) A survey of mobile
Deogirikar J, Vidhate A (2017) Security attacks inIoT: a survey. cloud computing: architecture, applications, and
In: International conference on I-SMAC (IoT in social, approaches. Wirel Commun Mob Comput 13:1587–1611
mobile, ana lytics and cloud), India Doswell W, Braxter B, Dabbs A, Nilsen W, Klem M (2013)
Devarajan M, Subramaniyaswamy V, Vijayakumar V (2019) mHealth: technology for nursing practice, education, and
Fog assisted personalized healthcare-support system for research. J Nurs Educ Pract 3(10):99–109
remote patients with diabetes. J Ambient Intell Human Elazhary H (2019) Internet of Things (IoT), mobile cloud,
Comput 10(1):3747–3760 cloudlet, mobile IoT, IoT cloud, fog, mobile edge, and
Díaz M, Martín C, Rubio B (2016) State-of-the-art challenges edge emerging computing paradigms: disambiguation
and open issues in the integration of internet of things and research directions. J Netw Comput Appl
and cloud comput ing. J Netw Comput Appl 67:99–117 128(15):105–140
Elazhary H, Althubyani A, Ahmed L, Alharbi B, Alzahrani N, task ofoading for time-sensitive applications in fog
Almutairi R (2017a) Context management for supporting computing. IEEE Syst J 13(3):2930–2941
con text-aware android applications development. Int J Jin D, Hannon C, Li Z, Cortes P, Ramaraju S, Burgess P, Buch
Interact Mob Technol 11(4):186–201 N, Sha hidehpour M (2016) Smart street lighting system: a
Elazhary H, Aloraini S, Aljuraid R (2017b) Context-aware platform for innovative smart city applications and a new
mobile application task ofoading to the cloud. Int J Adv frontier for cyber security. Electr J 29(10):28–35
Comput Sci Appl 8(5):381–390 Jin Y, Liu Y, Si W (2019) Intelligent agent distributed signal
Farahzadi A, Shams P, Rezazadeh J, Farahbakhsh R (2018) process ing for IoT. J Ambient Intell Human Comput,
Middleware technologies for cloud of things: a survey. Editorial for the special issue on
Digit Commun Netw 4(3):176–188 Karakikes I, Nathanail E (2017) Simulation techniques for
Gavalas D, Konstantopoulos C, Mastakas K, Pantziou G (2014) evaluating smart logistics solutions for sustainable urban
Mobile recommender systems in tourism. J Netw Comput distribution. Proc Eng 178:569–578
Appl 39:319–333 Karkar AG, Alja’am JM, Mahmood A (2017) Illustrate it! an Ara
Ghanbari Z, Navimipour N, Hosseinzadeh M, Darwesh A (2019) bic multimedia text-to-picture m-learning system. IEEE
Resource allocation mechanisms and approaches on the Access 5:12777–12787
Internet of Things. Clust Comput 22(4):1253–1282 Kaur RP, Kaur A (2014) Perspectives of mobile cloud comput
Gloukhovtsev M (2018) IOT security: challenges, solutions & ing: architecture, applications and issues. Int J Comput
future prospects. Knowledge sharing article. 2018 EMC Appl 101(3):9–14
Proven profes sional knowledge sharing Kekgathetse MB, Letsholo LK (2016) A survey on database
Gope P, Sikdar B (2019) Lightweight and privacy-preserving synchro nization algorithms for mobile device. J Theor
two-fac tor authentication scheme for IoT devices. IEEE Appl Inf Technol 86(1):1–9
Internet Things J 6(1):580–589 Kistler JJ, Satyanarayanan M (1992) Disconnected operation in
Gope P, Das AK, Kumar N, Cheng Y (2019) Lightweight and the coda fle system. ACM SIGOPS Oper Syst Rev 10(5):3–25
physi cally secure anonymous mutual authentication Korczak J, Kijewska K (2019) Smart logistics in the
protocol for real time data access in industrial wireless development of smart cities. Transp Res Proc 39:201–211
sensor networks. IEEE Trans Ind Inf 15(9):4957–4968 Krishnamurthy J (2016) JAMScript - a programming framework
Govindaraj P, Jaisankar N (2017) A review on various trust for cloud of things. M.Sc. thesis, School of Computer
models in cloud environment. J Eng Sci Technol Rev Science, McGill University, Montreal, Canada
10(2):213–219 Grifn PH (2017) Secure authentication on the Lee K, Kim J (2013) A mobile-based learning tool to improve
internet of things. In: writing skills of EFL learners. Proc Soc Behav Sci 106:112–119
Proceedings of SoutheastCon 2017, Charlotte, NC, USA, Lee K, Murray D, Hughes D, Joosen W (2010) Extending
pp 1–5. https://doi.org/10.1109/SECON.2017.7925274 sensor net works into the cloud using Amazon Web services. In
Harari G, Müller S, Aung M, Rentfrow P (2017) Smartphone Proc. IEEE Int. Conf. Netw. Embedded Syst. Enterprise Appl.
sensing methods for studying behavior in everyday life. (NESEA), pp 1–7
Curr Opin Behav Sci 18:83–90 Li W, Santos I, Delicato FC, Pires PF, Pirmez L, Wei W, Song
He W, Yan G, Xu LD (2014) Developing vehicular data cloud H, Zomaya A, Khan S (2017) System modelling and
services in the IoT environment. IEEE Trans Ind Inf performance evaluation of a three-tier cloud of things.
10(2):1587–1595 Heiler S (1995) Semantic interoperability. Future Gener Comput Syst 70:104–125
ACM Computing Surveys Hologram (2019) https://hologram.io/. Machado CG, Winroth MP, daSilva EHDR (2020) Sustainable
Accessed 10 Dec 2019 Horvath AS, Agrawal R (2015) Trust in manu facturing in industry 4.0: an emerging research
cloud computing. In: Proceed igns of SoutheastCon 2015, Fort agenda. Int J Prod Res 58(5):1462–1484
Lauderdale, FL, pp 1–8 Mahmoud M, Rodrigues J, Saleem K, Al-Muhtadi J, Kumar N,
Hussain W, Hussain FK, Hussain OK (2014) Maintaining trust in Koro taevf V (2018) Towards energy-aware fog-enabled cloud
cloud computing through SLA monitoring. In: Loo CK, of things for healthcare. Comput Electr Eng 67:58–69
Yap KS, Wong KW, Beng Jin AT, Huang K (eds) Neural Marudhadevi D, Dhatchayani VN, Sriram VSS (2015) A trust
information process ing. ICONIP 2014. Lecture Notes in evalua tion model for cloud computing using service level
Computer Science 8836. Springer, Cham agreement. Comput J 58(10):2225–2232
Iwan S, Małecki K (2016) Data fows in urban freight transport Medina J, Espinilla M, García-Fernández Á, Martínez L (2018)
man agement system. In: Communications in computer Intel ligent multi-dose medication controller for fever: from
and informa tion science, Springer-Verlag, Berlin, wearable devices to remote dispensers. Comput Electr Eng
Heidelberg, vol 531,pp 1–10 65:400–412
Jabeur N et al (2017) Toward leveraging smart logistics Microsoft Azure (2019)
collabora tion with a multi-agent system based solution. https://azure.microsoft.com/en-us/features/ iot-
Proc Comput Sci 109:672–679 accelerators/. Accessed 9 Dec 2019
Jiang Y-L, Chen Y-S, Yang S-W, Wu C-H (2019) Energy-efcient

13
Cloud of Things: architecture, applications and challenges

Mosa A, Yoo I, Sheets L (2012) A systematic review of Oludele A, Oluwabukola O (2016) A survey of mobile cloud
healthcare appli cations for smartphones. BMC Med Inform computing applications: perspectives and challenges. In:
Decis Making 12:67 MuleSoft (2019) International multi conference on complexity, informatics, and
https://www.mulesoft.com/integration-solutions/api/iot. cybernetics
Accessed 10 Dec 2019 Ozinga S, Linder S, Alberts J (2017) Use of mobile device
Munir K (2018) Advancing consumer centric fog computing accelerometry to enhance evaluation of postural
architec tures. A volume in the advances in computer and instability in Parkinson’s disease. Arch Phys Med Rehabil
electrical engi neering (ACEE) book series, IGI Global 98(4):649–658
Narwane VS, Narkhede BE, Gardas BB, Raut RD (2019) Cloud Particle (2019) https://www.particle.io/. Accessed 9 Dec 2019
manufac turing issues and its adoption: past, present, and Patel YS, Parmar T (2016) Cloud of things: a state-of-the-art
future. Int J Manag Concepts Philos 12(2):168–199 review on integration of internet of things with cloud computing.
In: Proceed ings of national conference on contemporary on mobile cloud computing. Int J Comput Appl 113(16):4–9
computing (NC3– 2016), Int J Comput Appl, pp 37–41 Sharma AK, Soni P (2013) Mobile cloud computing (MCC):
Pattama C (2018) Smart logistic system by IOT technology. In: open research issues. Int J Innov Eng Technol 2(1):24–27
Proceed ings of the 6th international conference on information Shrivastava R, Hota C, Shrivastava P (2017) Protection against
and education technology (ICIET ’18), ACM, New York, NY, code exploitation using ROP and check-summing in IoT
USA, pp 149–153 environment. In: Fifth international conference on
Perez M, Kumar S (2017) A quick survey on cloud computing information and communica tion technology, p 6
and associated security, mobility and IoT issues. J SigFox (2019) https://sigfox.com/. Accessed 10 Dec 2019
Comput Commun 5(12):80–95 Somula R, Sasikala R (2018) A survey on mobile cloud
Qi H, Gani A (2012) Research on mobile cloud computing: computing: mobile computing and cloud computing (MCC =
review, trend and perspectives. In: Proceeding of second MC + CC). Scalable Comput Pract Exp 19(4):309–337
international confer ence on digital information and Sprute D, Tönnies K, König M (2019) Interactive restriction of a
communication technology and its applications (DICTAP), mobile robot’s workspace in a smart home environment. J
Malaysia, pp 195–202 Ambient Intell Smart Environ 11(6):475–494
Qu Z, Zhang G, Cao H, Xie J (2017) leo satellite constellation Stergiou K, Psannis E, Kim BG, Gupta B (2018) Secure
for internet of things. IEEE Access 5:18391–18401 integration of IoT and cloud computing. Future Gener Comput
Rahman A, Sultan K, Dash S, Khan MAA (2018) Management Syst 78:964–975 Suciu G, Vulpe A, Halunga S, Fratu O,
of resource usage in mobile cloud computing. Int J Pure Todoran G, Suciu V (2013) Smart cities built on resilient cloud
Appl Math 119(16):255–261 computing and secure internet of things. In: 19th international
Rahman A et al. (2018b) A comprehensive study of mobile conference on control systems and computer science (CSCS),
computing in telemedicine. In: Proceedings of second 2013, IEEE, pp 513–518
international conference, ICAICR 2018, Shimla, India, Tanaka S, Fujishima K, Mimura N, Ohashi T, Tanaka M (2016)
14–15 July 2018 iot sys tem security issues and solution approaches. In:
Rajak N, Shukla D (2018) Comparative studly of cloud Featured Articles II, vol 65, no 8, p 5
computing and mobile cloud computing. Int J Eng Sci Res ThingSpeak (2019) https://thingspeak.com/. Accessed 9 Dec
Technol 7(3):734–739 Rawahi ASA, Lee K, Robinson J, Lotf A 2019 Ujager FS, Mahmood A, Khatoon S (2019) Wellness
(2018) An evaluation of optimisa tion approaches in cloud of determination of the elderly using spatio-temporal correlation
things resource trading. In: Proc. IEEE 6th int. conf. future analysis of daily activities. J Ambient Intell Smart Environ
internet things cloud (FiCloud), pp 208–215 Razvi SM, Abdullah 11(6):515–526
D, Mznah R, Riman AS (2015) IoT cloud-sensor secure Uma K, Parvin NA, Poorani ARM (2017) Analysis of various
architecture for smart home. In: Proceedings of int’l conf. issues in cloud of things (COT). J Chem Pharm Sci 10(1):532–
security and management (SAM’15) 535 Vinh TL (2018) Security and trust in mobile cloud
Renner T, Kliem A, Kao O (2014) The device cloud -applying computing. PhD Thesis
cloud com puting concepts to the internet of things. In Vishwakarma G, Lee W (2018) Exploiting JTAG and its
Proceedings of IEEE 11th intl conf on ubiquitous intelligence & mitigation in IOT: a survey. Future Internet 10(121):18
computing Wang Q, Deters R (2009) SOA’s last mile-connecting
Rimale Z, Benlahmar E, Tragha A, Guemmat KE (2016) Survey smartphones to the service cloud. In: Proceedings of
on the use of the mobile learning based on mobile cloud 2009 IEEE international confer ence on cloud computing,
computing. iJIM 10(3):35–41 Bangalore, pp 80–87
Roman R, López J, Mambo M (2018) Mobile edge computing, Witkowski K (2017) Internet of things, big data, industrie 4.0 –
fog et al.: a survey and analysis of security threats and innova tive solutions in logistics and supply chains
challenges. Future Gener Comput Syst 78:680–698 management. Proc Eng 182:763–769
Rountree D, Castrillo I (2014) The basics of cloud computing: Wu X (2018) Study on trust model for multi-users in cloud
under standing the fundamentals of cloud computing in computing. Int J Netw Secur 20(4):674–682
theory and prac tice, 1st edn. Elsevier, Amsterdam Yao D, Yu C, Jin H, Zhou J (2013) Energy efcient task
Sahmim S, Gharsellaoui H (2017) Privacy and security in scheduling in mobile cloud computing. In: Network and
internet-based computing: cloud computing, internet of parallel computing, vol 8147. Springer, LNCS, pp 344–
things, cloud of things: a review. In: Knowledge-based 355
and intelligent information & engi neering systems: Yang W, Hwang S (2013) iTravel: a recommender system in
proceedings of the 21st international conference mobile peer to-peer environment. J Syst Softw 86:12–20
marseille, France, pp 1516–1522 Yasir S, Crespi N, Rehmani MH, Copeland R (2016)
SalesForce (2019) https://www.salesforce.com/products/feld- Exploitation of social IoT for recommendation services. In:
service/ overview/. Accessed 9 Dec 2019 Proceedings of 2016 IEEE 3rd World forum on internet of
Sanabria AR, Kelsey T, Dobson S, Ye J (2019) Representation things (WF-IoT), Reston, VA, pp 359–364.
learning for minority and subtle activities in a smart home https://doi.org/10.1109/WF-IoT.2016.7845500
environment. J Ambient Intell Smart Environ 11(6):495– Zaslavsky A, Perera C, Georgakopoulos D (2013) Sensing as a
513 service and big data. In: Proceedings of the international
Sandoval RM, Carrasco SC, Sanchez AAG, Haro JG (2019) A conference on advances in cloud computing (ACC),
reinforce ment learning-based framework for the Bangalore, India, ArXiv Pre print arXiv:1301.0159
exploitation of multiple rats in the IoT. IEEE Access
7:123341–123354. https://doi.org/10.1109/ Publisher’s Note Springer Nature remains neutral with regard
ACCESS.2019.2938084 to jurisdictional claims in published maps and institutional
Shamim SM, Sarker A, Bahar AN, Rahman MA (2015) A review afliations.

13