Hindawi Publishing Corporation

International Journal of Distributed Sensor Networks

Volume 2013, Article ID 917923, 18 pages
http://dx.doi.org/10.1155/2013/917923

Review Article
A Survey on Sensor-Cloud: Architecture,
Applications, and Approaches

Atif Alamri, Wasai Shadab Ansari, Mohammad Mehedi Hassan,

M. Shamim Hossain, Abdulhameed Alelaiwi, and M. Anwar Hossain
College of Computer and Information Sciences, King Saud University, P.O. Box 51178, Riyadh 11543, Saudi Arabia

Correspondence should be addressed to Atif Alamri; [email protected]

Received 25 May 2012; Revised 30 September 2012; Accepted 28 November 2012

Academic Editor: Damla Turgut

Copyright © 2013 Atif Alamri et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Nowadays, wireless sensor network (WSN) applications have been used in several important areas, such as healthcare, military,
critical infrastructure monitoring, environment monitoring, and manufacturing. However, due to the limitations of WSNs in
terms of memory, energy, computation, communication, and scalability, efficient management of the large number of WSNs data
in these areas is an important issue to deal with. There is a need for a powerful and scalable high-performance computing and
massive storage infrastructure for real-time processing and storing of the WSN data as well as analysis (online and offline) of
the processed information under context using inherently complex models to extract events of interest. In this scenario, cloud
computing is becoming a promising technology to provide a flexible stack of massive computing, storage, and software services in
a scalable and virtualized manner at low cost. Therefore, in recent years, Sensor-Cloud infrastructure is becoming popular that can
provide an open, flexible, and reconfigurable platform for several monitoring and controlling applications. In this paper, we present
a comprehensive study of representative works on Sensor-Cloud infrastructure, which will provide general readers an overview of
the Sensor-Cloud platform including its definition, architecture, and applications. The research challenges, existing solutions, and
approaches as well as future research directions are also discussed in this paper.

1. Introduction as sensor nodes are camera sensor, accelerometer sensor,

thermal sensor, microphone sensor, and so forth.
The advancement and application of wireless sensor networks Currently, WSNs are being utilized in several areas
become an invincible trend into the various industrial, envi- like healthcare, defense such as military target tracking
ronmental, and commercial fields. A typical sensor network and surveillance [4, 5], government and environmental ser-
may consist of a number of sensor nodes acting upon together vices like natural disaster relief [6], hazardous environment
to monitor a region and fetch data about the surroundings. A exploration, and seismic sensing [7], and so forth. These
WSN contains spatially distributed self-regulated sensors that sensors may provide various useful data when they are
can cooperatively monitor the environmental conditions, like closely attached to each of their respective applications and
sound, temperature, pressure, motion, vibration, pollution, services directly [8]. However, sensor networks have to face
and so forth [1, 2]. Each node in a sensor network is loaded many issues and challenges regarding their communications
with a radio transceiver or some other wireless communica- (like short communication range, security and privacy, reli-
tion device, a small microcontroller, and an energy source ability, mobility, etc.) and resources (like power consider-
most often cells/battery. The nodes of sensor network have ations, storage capacity, processing capabilities, bandwidth
cooperative capabilities, which are usually deployed in a availability, etc.). Besides, WSN has its own resource and
random manner. These sensor nodes basically consist of three design constraints. Design constraints are application specific
parts: sensing, processing, and communicating [3]. Some of and dependent on monitored environment. Based on the
the most common sensor devices deployed in sensor network monitored environment, network size in WSN varies. For
2 International Journal of Distributed Sensor Networks

monitoring a small area, fewer nodes are required to form a information from the sensor nodes of WSN, compresses it,
network whereas the coverage of a very large area requires and transmits it back to the cloud gateway which in turn
a huge number of sensor nodes. For monitoring large envi- decompresses it and stores it in the cloud storage server,
ronment, there is limited communication between nodes due which is sufficiently large [13].
to obstructions into the environment, which in turn affects Sensor-Cloud can be used in many real-life applications
the overall network topology (or connectivity) [9]. All these like environmental monitoring, disaster monitoring, tele-
limitations on sensor networks would probably impede the metric, agriculture, irrigation, healthcare, and so forth. As
service performance and quality. In the midst of these issues, an illustration, we can use the Sensor-Cloud infrastructure
the emergence of cloud computing is seen as a remedy. for deploying health-related applications such as monitoring
Cloud computing has been evolved as the future genera- patients with cardiovascular disease, blood sugar followup,
tion’s computing paradigm. The US NIST (National Institute sleep activity pattern monitoring, diabetics monitoring, and
of Standards and Technology) defines the concept of Cloud so forth. In traditional approach, the trials of individual’s data
computing as follows: like level of blood sugar, weight, heart rate, pulse rate, and
so forth are reported everyday through some telemedicine
Cloud computing is a model for enabling con-
interface [14]. The patient’s trial information is sent to a
venient, on demand network access to a shared
dedicated server and is stored there for doctors or caregivers
pool of configurable computing resources (e.g., net-
to analyze it sometime later. This system suffers from a level of
works, servers, storage, applications, and services)
adversity when the patient randomly moves from its current
that can be rapidly provisioned and released with
location, that is, when a patient is “on the go.” Thus, a more
minimal management effort or service provider
progressive, rapid, and mobile approach is needed where
interaction [5].
the recorded data from several sensor nodes of a WSN can
Cloud computing allows the systems and users to use be processed in pipelined and parallel fashion, and thereby
Platform as a Service (PaaS), for example, Operating Systems to make the system easier to scale and be cost-effective in
(OSs), Infrastructure as a Service (IaaS), for example, storages terms of resources available. The pipeline processing of data
and servers, and Software as a Service (SaaS), for example, sets or instructions enable the overlapped operations into
application level programs, and so forth at a very low cost a conceptual pipe with all the stages of pipes processing
which are being provided by several cloud providers (e.g., simultaneously but handling of the sensor data stream is not
example Amazon, Google, and Microsoft) on the basis of pay that straight forward and will be dependent on the nature of
per use services [10]. Cloud computing platform dynamically the algorithm [15]. The integration of Sensor-Cloud can serve
provisions, configures, and reconfigures the servers as and as a remedy in this direction.
when needed by end users. These servers can be in the form This paper presents a comprehensive survey on Sensor-
of virtual machines or physical machines in the cloud. Cloud Cloud. Section 2 of this paper provides a brief overview of
computing renders the two major trends in IT: (1) efficiency, this concept, including its definition, architecture, and its
which is achieved through the highly scalable hardware and advantage. Section 3 discusses the service life cycle model
software resources, and (2) agility, which is achieved through and layered approach of Sensor-Cloud architecture. Section 4
parallel batch processing, using computer-intensive business provides the Sensor-Cloud service creation and renovation
analytics and real-time mobile interactive applications that capabilities. Section 5 presents the multiservice provisioning
respond to user requirements [11]. The benefits of cloud on multiple platforms. Section 6 discusses the technical
computing are that the end users need not to worry about the comparison of different approaches in sensor-cloud infras-
exact location of servers and switch to their application by tructure. Next, Section 7 presents several issues that may arise
connecting to the server on cloud and start working without in Sensor-Cloud and some approaches to address these issues.
any hassle. Finally, we summarize and conclude the survey in Section 8.
Sensor-Cloud infrastructure [12] has been evolved and
proposed by several IT people in the present days. Sensor- 2. Overview and Related Works
Cloud infrastructure is the extended form of cloud comput-
ing to manage the sensors which are scattered throughout Sensor-Cloud is a new paradigm for cloud computing that
the network (WSN). Due to the increasing demand of sensor uses the physical sensors to accumulate its data and transmit
network applications and their support in cloud computing all sensor data into a cloud computing infrastructure. Sensor-
for a number of services, Sensor-Cloud service architecture Cloud handles sensor data efficiently, which is used for many
is introduced as an integration of cloud computing into monitoring applications.
the WSN to innovate a number of other new services.
When WSN is integrated with cloud computing environment, 2.1. What is a Sensor-Cloud? According to IntelliSys, Sensor-
several shortfalls of WSN like storage capacity of the data Cloud can be defined as follows:
collected on sensor nodes and processing of these data
together would become much easier. Since cloud computing An infrastructure that allows truly pervasive com-
provides a vast storage capacity and processing capabilities, it putation using sensors as an interface between
enables collecting the huge amount of sensor data by linking physical and cyber worlds, the data-compute clus-
the WSN and cloud through the gateways on both sides, that ters as the cyber backbone and the internet as the
is, sensor gateway and cloud gateway. Sensor gateway collects communication medium [16, 17].
International Journal of Distributed Sensor Networks 3

According to MicroStrains’s Sensor-Cloud definition “it (2) Network-Centric Healthcare-QoS refers to a number
is a unique sensor data storage, visualization and remote of patients supported and message throughput.
management platform that leverage [sic] powerful cloud com- (3) Healthcare professional-centric Healthcare-QoS refers
puting technologies to provide excellent data scalability, rapid to cognitive load of healthcare professionals and the
visualization, and user programmable analysis. It is origi- number of correct medical decisions.
nally designed to support long-term deployments of MicroS-
train wireless sensors, Sensor-Cloud now supports any web- Sensor modeling language (SML) [24] can be used to
connected third party device, sensor, or sensor network through represent any physical sensor’s metadata like their type,
a simple OpenData API” [18]. accuracy, their physical location, and so forth. It also uses
A Sensor-Cloud collects and processes information from XML encoding for the measurement and description pro-
several sensor networks, enables information sharing on cesses of physical sensors. This XML encoding for physical
big scale, and collaborates with the applications on cloud sensors enabled these to be implemented across several
among users. It integrates several networks with a number different hardware, platforms (OS), applications, and so forth
of sensing applications and cloud computing platform by with relatively less human intervention. To transliterate the
allowing applications to be cross-disciplinary that may span commands coming from users to virtual sensors and in
over multiple organizations [17]. Sensor-Cloud enables users turn to the commands for their pertinent physical sensors,
to easily gather, access, process, visualize, analyze, store, a mapping is done between the physical and virtual sensors.
share, and search for a large number of sensor data from
several types of applications and by using the computational 2.2. Architecture of Sensor-Cloud. Cloud computing service
IT and storage resources of the cloud. framework delivers the services of shared network through
In a sensor network, the sensors are utilized by their which the users are benefited by the services, and they are
specific application for a special purpose, and this application not concerned with the implementation details of the services
handles both the sensor data and the sensor itself such that provided to them. When a user requests, the service instances
other applications cannot use this. This makes wastage of (e.g., virtual sensors) generated by cloud computing services
valuable sensor resources that may be effectively utilized are automatically provisioned to them [12, 19].
when integrating with other application’s infrastructure. To Some previous studies on physical sensors focused on
realize this scenario, Sensor-Cloud infrastructure is used that routing [25], clock synchronization [26], data processing
enables the sensors to be utilized on an IT infrastructure [27], power management [28], OS [29], localization [30],
by virtualizing the physical sensor on a cloud computing and programming [31]. However, few studies concentrate on
platform. These virtualized sensors on a cloud computing physical sensor management because these physical sensors
platform are dynamic in nature and hence facilitate automatic are bound closely to their specific application as well as to
provisioning of its services as and when required by users its tangible users directly. However, users, other than their
[19]. Furthermore, users need not to worry about the phys- relevant sensor services, cannot use these physical sensors
ical locations of multiple physical sensors and the gapping directly when needed. Therefore, these physical sensors
between physical sensors; instead, they can supervise these should be supervised by some special sensor-management
virtual sensors using some standard functions [12]. schemes. The Sensor-Cloud infrastructure would subsidize
Within the Sensor-Cloud infrastructure, to obtain QoS, the sensor system management, which ensures that the data-
the virtual sensors are monitored regularly so users can management usability of sensor resources would be fairly
destroy their virtual sensors when they becomes meaningless improved.
[20]. A user interface is provisioned by this Sensor-Cloud There exists no application that can make use of every
infrastructure for administering, that is, for controlling or kind of physical sensors at all times; instead, each application
monitoring the virtual sensors, provisioning and destroying required pertinent physical sensors for its fulfillment. To
virtual sensors, registering and deleting of physical sensors, realize this concept, publish/subscription [32] mechanism is
and for admitting the deleting users. For example, in a being employed for choosing the appropriate physical sensor
health monitoring environment, a patient may use a wearable [33]. In multiple sensor networks, every sensor network
computing system (that may include wearable accelerometer publishes its sensor data and metadata. The metadata com-
sensors, proximity sensors, temperature sensors, etc.) like Life prises of the types, locations, and so forth for the physical
Shirt [21] and Smart Shirt [22] or may use a handheld device sensors. Application either subscribes to one or maybe to
loaded with sensors, and consequently the data captured by more sensor networks to retrieve real-time data from the
the sensors may be made accessible to the doctors. But out physical sensors by allowing each application to opt for the
of these computing systems, active continuous monitoring appropriate physical sensors’ type. The Sensor-Cloud infras-
is most demanding, and it involves the patient wearing tructure procreates virtual sensors from multiple physical
monitoring devices to obtain pervasive coverage without sensors, which can then be utilized by users. However, prior
being inputted or intervened [23]. However, the diverse to availing the virtual sensor facility, users should probe first
monitoring scheme defers in their QoS requirements, which for the physical sensors’ availability and might also inspect the
are as follows. physical sensors’ faults to maintain the data quality emerging
from these physical sensors. Also on every sensor node,
(1) Patient-Centric Healthcare-QoS refers to monitoring application program senses the application and sends the
delay and reliability of message delivery. sensor data back to the gateway in the cloud directly through
4 International Journal of Distributed Sensor Networks

the base station or in multihop manner through other nodes groups (virtual sensors). Once templates are prepared, the
[5, 13]. virtual sensors are able to share the related and contiguous
Sensor-Cloud infrastructure provides service instances physical sensors to receive quality sensor data. Users then
(virtual sensors) automatically to the end users as and when request these virtual sensors by choosing the appropriate
requested, in such a way that these virtual sensors are part service templates, use their service instances (virtual sensors)
of their IT resources (like disk storage, CPU, memory, etc.) after being provisioned, and discharge them when became
[34]. These service instances and their associated appropriate useless [12].
sensor data can be used by the end users via a user interface
through the web crawlers as described in Figure 1. However, 2.3. Advantages of Sensor-Cloud. Cloud computing is very
for the service instance generation, the IT resources (like encouraging solution for Sensor-Cloud infrastructure due
CPU, storage devices, etc.), sensor capable devices, and ser- to several reasons like the agility, reliability, portability,
vice templates (which is used to create virtual sensors) should real-time, flexibility, and so forth. Structural health and
be prepared first in Sensor-Cloud infrastructure. Users make environment-based monitoring contains highly sensitive
the request for service instances according to their needs by data and applications of these types cannot be handled
selecting an appropriate service template of Sensor-Cloud, by normal data tools available in terms of data scalability,
which will then provide the needed service instances freely performance, programmability, or accessibility. So a better
and automatically because of cloud computing services inte- infrastructure is needed that may contain tools to cope
gration. Once service instances become useless, they can then with these highly sensitive applications in real time. In the
be deleted quickly by users to avoid the utilization charges following, we describe the several advantages and benefits of
for these resources. Sensor service provider will manage the Sensor-Cloud infrastructure that may be the cause of its glory,
service templates (ST) and it can add or delete the new service and these are as follows.
template when the required template is no longer needed by
applications and services [12]. Automation of services played (1) Analysis. The integration of huge accumulated sensor
a vital role in provisioning of cloud computing services, and data from several sensor networks and the cloud
automation can cause the delivery time of services to be better computing model make it attractive for various kinds
[19]. Before the emergence of cloud computing, services of analyses required by users through provisioning of
were provided by human influence and the performance the scalable processing power [34].
metrics like efficiency, flexibility, delivery times, and so forth
would have experienced an adverse effect on the system. (2) Scalability. Sensor-Cloud enables the earlier sensor
However, the cloud computing service model has reduced the networks to scale on very large size because of the
cost expenses and delivery time and has also improved the large routing architecture of cloud [16]. It means that
efficiency and flexibility. as the need for resources increases, organizations can
The physical sensors are ranked on a basis of their sensor scale or add the extra services from cloud computing
readings as well as on their actual distance from an event. vendors without having to invest heavily for these
The authors of [35] proposed a technique (FIND) to locate additional hardware resources [36].
physical sensors having data faults by assuming a mismatch (3) Collaboration. Sensor-Cloud enables the huge sensor
between the distance rank and sensor data rank. However, data to be shared by different groups of consumers
the study led by FIND aims at the assessment of physical through collaboration of various physical sensor net-
sensors faults, and there is a close relation between the virtual works [16]. It eases the collaboration among several
and physical sensors and hence a virtual sensor will provide users and applications for huge data sharing on the
incorrect results if their relevant physical sensors are faulty. cloud.
It concludes that the virtual and physical sensors cooperate
while delivering the sensor data report to its applications, (4) Visualization. Sensor-Cloud platform provide a visu-
and to maintain the best report both the virtual and physical alization API to be used for representing the diagrams
sensors should be faultless. with the stored and retrieved sensor data from several
Since the cloud computing enables the physical sensors device assets. Through the visualization tools, users
to be virtualized, the users of the Sensor-Cloud infrastructure can predict the possible future trends that have to be
need not to worry about the status of their connected physical incurred [37].
sensors (i.e., whether a fault free or not). However, they
(5) Free Provisioning of Increased Data storage and Pro-
should concern only with the status of their virtual sensors
cessing Power. It provides free data storage and orga-
provided only when the users are not concerned with the
nizations may put their data rather than putting onto
accurate results. To achieve accurate results, users must
private computer systems without hassle. It provides
be concerned about the status of physical sensors too. In
enormous processing facility and storage resources to
a Sensor-Cloud infrastructure, sensors owners are free to
handle data of large-scale applications [16, 38].
register or unregister their physical sensors and can join this
infrastructure. These IT resources (physical sensors, database (6) Dynamic Provisioning of Services. Users of Sensor-
servers, processors, etc.) and sensor devices are then prepared Cloud can access their relevant information from
to become operational. After that, templates are created for wherever they want and whenever they need rather
generating the service instances (virtual sensors) and its than being stick to their desks [38].
International Journal of Distributed Sensor Networks 5

Requesting the
End users service instances
Application 1 or (virtual sensor)

Application 2

Application“ 𝑛”
Sensor-Cloud owner or
administrator

Sensor cloud infrastructure

Template Sensor management service
Prepare service
templates Repository
Virtual sensor group

ST-A
Service A

ST-B
Managing
Service B
infrastructure
Virtual sensor or
service instances
Registry
Gateway
Other IT
resources Gateway

Physical sensors 2

Physical sensors 1
Registering

1 2 3 ··· 𝑛

Sensor owner

Figure 1: Brief overview of Sensor-Cloud architecture.

(7) Multitenancy. The number of services from several (10) Agility of Services. Sensor-Cloud provides agile ser-
service providers can be integrated easily through vices and the users can provision the expensive
cloud and Internet for numerous service innovations technological infrastructure resources with less cost
to meet user’s demand [17]. Sensor-Cloud allows the [37]. The integration of wireless sensor networks with
accessibility to several numbers of data centers placed cloud allows the high-speed processing of data using
anywhere on the network world [36]. immense processing capability of cloud.
(8) Automation. Automation played a vital role in (11) Resource Optimization. Sensor-Cloud infrastructure
provisioning of Sensor-Cloud computing services. enables the resource optimization by allowing the
Automation of services improved the delivery time to sharing of resources for several number of appli-
a great extent [19]. cations [37]. The integration of sensors with cloud
(9) Flexibility. Sensor-Cloud provides more flexibility to enables gradual reduction of resource cost and
its users than the past computing methods. It provides achieves higher gains of services. With Sensor-Cloud,
flexibility to use random applications in any number both the small and midsized organizations can benefit
of times and allows sharing of sensor resources under from an enormous resource infrastructure without
flexible usage environment [16]. having to involve and administer it directly [36].
6 International Journal of Distributed Sensor Networks

Preparing
IT resources
and sensor
devices
Unregistering Preparing
the service service
templates templates to
or physical create service
services instances

Service life cycle

development

Unregistering/ Users
requesting the
deleting the service
service
instances instances
Utilising the
service
instances

User interface via web

crawlers

Monitoring Automated provisioning Controlling

of data

Virtual sensor groups

Physical sensors 1 Physical sensors 2 Physical sensors 𝑛

Figure 2: Demonstration of Sensor-Cloud life cycle development phases.

(12) Quick Response Time. The integration of WSN’s with 3.1. Service Life Cycle Model of Sensor-Cloud. Before creating
cloud provides a very quick response to the user, that the service instances within Sensor-Cloud infrastructure,
is, in real-time due to the large routing architecture preparation phase [12] is needed, and this includes the
of cloud [39]. The quick response time of data feeds following.
from several sensor networks or devices allows users
(1) Preparing the IT resources (processors, storage, disk,
to make critical decisions in near real time.
memory etc.).
(2) Preparing the physical sensor devices.
3. Sensor-Cloud Service Life-Cycle Model and (3) Preparing the service templates.
Its Layered Structure
Figure 2 depicts the Sensor-Cloud service life cycle.
The service life-cycle model and the layered structure of The users of the sensors can select the appropriate service
Sensor-Cloud infrastructures are illustrated in the following. template and request the required service instances. These
International Journal of Distributed Sensor Networks 7

service instances are provided automatically and freely to the 4. Service Creation and
users, which can then be deleted quickly when they become Innovation Capabilities
useless. From a single service template, multiple numbers of
service instances can be created. Service provider regulates Sensors are very limited and specific to their applications or
the service templates and can add new service templates as services when they are linked to a typical sensor network.
and when required by a different number of users [8]. Therefore, the numbers of organizations that can provide the
sensor services are very limited. However, when the services
of sensors move onto the cloud, it is possible to include them
3.2. Layered Structure of Sensor-Cloud. Figure 3 depicts the to realize a variety of applications [8, 12]. A number of services
layered architecture of the Sensor-Cloud platform, which is can be provided to the users for different applications such
divided mainly into three layers: as health applications, environmental monitoring, industrial
tasks (e.g., refining), surveillance, senior residents monitor-
ing, or even the applications that monitor the vibration in
(1) user and application layers, buildings during an earthquake.
In the Sensor-Cloud infrastructure, the sensors and
(2) Sensor-Cloud and virtualization layers, service templates are constructed as catalog menu service
on the cloud, and the requesters can create new sensor
services with the existing sensors in these service instances.
(3) template creation and tangible sensors layers [8].
For example, service requester can create a sensor service to
analyze the impact of earthquake to each floor or room of
Layer 1. This layer deals with the users and their relevant the rehabilitation center or hospital, and at the same time
applications. Several users want to access the valuable sensor it can also create sensor services to support older residents
data from different OS platforms, such as mobile phones OS, with the same set of sensors (virtualized sensors). This service
Windows OS, or Mac OS for a variety of applications. This will then help the caregiver to shift the older adults one by
structure allows users of different platforms to access and one into a safe place. Using the identical sensor services for
utilize the sensor data without facing any problem because healthcare, another service requester can create dissimilar
of the high availability of cloud infrastructure and storage. sensor service to track the patient’s medicine intake and then
to analyze the effectiveness of pills through the use of some
Layer 2. This layer deals with virtualization of the physical selected healthcare sensors. Thus, the service requesters can
sensors and resources in the cloud. The virtualization enables be provided with new services using the same set of sensors
the provisioning of cloud-based sensor services and other IT on cloud service platforms. This will reduce the cost for
resources remotely to the end-user without being worried resource usage and could have numerous elastic merits to it.
about the sensors exact locations. The virtualized sensors In this section, several existing Sensor-Cloud applications are
are created by using the service templates automatically. described.
Service templates are prepared by the service providers as
service catalog, and this catalog enables the creation of service
instances automatically that are accessed by multiple users 4.1. Existing Sensor-Cloud Applications. There exist a number
[8, 19]. of services based on Sensor-Cloud infrastructure to store
and process the sensor-based information. Few of them are
Layer 3. This is the last layer which deals with the service described briefly as below.
template creation and service catalog definition layers in
forming catalog menu. Physical sensors are located and
retrieved from this layer. Since each physical sensor has 4.1.1. Nimbits. Nimbit [40] is a free and social service that
its own control and data collection mechanism, standard is used to record and share sensor data on cloud. It is a
mechanisms are defined and used to access sensors without cloud-based data processing service and is an open-source
concerning the differences among various physical sensors platform for the IoT (Internet of Things). We can feed the
[12]. Standard functions are defined to access the virtual versatile numeric, text-based, JSON, GPS, or XML values
sensors by the users. In Layer 2, Sensor-Cloud infrastructure by creating a data point in the cloud. The data points can
translates standard functions of virtual sensors into some be connected to Scalable Vector Graphic (SVG) process
specific functions for diverse physical sensors in Layer 3. control, spreadsheets, diagrams, websites, and more. Data
Physical sensors are XML encoded that enable the services points can also be configured to generate alerts data-relay
provided through these sensors to be utilized on various to social networks and to perform calculations. Nimbits also
platforms without being worried to convert them onto several provide an alert management mechanism, data compression
platforms [8]. mechanism, and data calculation on received sensor data by
Sensor-Cloud provides a web-based aggregation platform employing some simple mathematical formulas.
for sensor data that is flexible enough to help in developing
user-based applications. It allows users quick development 4.1.2. Pachube Platform. Pachube [41] is one of the first online
and deployment of data processing applications and gives database service providers, which allows us to connect sensor
programming language flexibility in accordance to their data to the web. It is a real-time cloud-based platform for
needs [18]. IoT with a scalable infrastructure that enables us to configure
8 International Journal of Distributed Sensor Networks

User
OS 1 application 1
User
OS 2 application 2
OS 3 User
application 3
OS “𝑁” User
application “n”

User interface
Layer 1

Layer 2 Sensor management

into the cloud
Virtual
sensor VSG 2 VSG “𝑁”
Virtual group (VSG) 1
sensor
(VS)/ VS/SI VS/SI VS/SI VS/SI VS/SI “ 𝑛 ”
service
instances
(SI)
···
Layer 3
Service Service Service Service Service
templates templates templates templates templates “ 𝑛 ”
XML encoded XML encoded XML encoded XML encoded XML encoded
physical physical physical physical physical
sensors sensors sensors sensors sensors “𝑛 ”

Figure 3: Layered structure of Sensor-Cloud model.

IoT products and services, store, share, and discover real- assets with proper update of its status can be created. ThingS-
time energy, environment, and healthcare sensor data from peak allows numeric data processing like averaging, time-
devices and buildings around us. Pachube has a very inter- scaling, rounding, median, summing, and so forth to store
active website for managing the sensor data and an open and retrieve the numeric and alphanumeric data. ThingSpeak
easily-accessible API. Pachube system provides free usage application features a JavaScript-based charts, read/write API
and has several numbers of interfaces for producing a sensor key management, and a time-zone management.
or mobile-based applications for managing the sensor data Although the above services are able to visualize the
within a cloud framework anytime. sensor data and sensor-driven information, they are lacking
secure access to data and interface availability for linking
the external or mobile applications for further processing.
4.1.3. IDigi. iDigi [42] is a machine-to-machine (M2M) It means that most of these aforementioned projects do not
platform as a service PaaS that minimizes the barriers to address the issues of data management and interoperability
build scalable, secure, and cost-effective solutions, which can issues caused by heterogeneous data resources found in
bind the enterprise applications and device assets together. the present modern environmental tracking or electronic
iDigi eases the connectivity of remote assets devices and healthcare systems. But introducing these aforementioned
provides all the tools to manage, store, connect, and move the works with Cloud computing infrastructure may overcome
information across the enterprise irrespective of its reache. the issues related to heterogeneous data access and data
To simplify the remote device connectivity and integration, it management functionality [37].
uses connector software called iDigi Dia. Regardless of the
network location, iDigi platform manages communication
4.2. Emerging Sensor-Cloud Applications. There are many
between remote device assets and enterprise applications.
other applications that are emerging based on the Sensor-
Cloud infrastructure, which can be summarized as follows.
4.1.4. ThingSpeak. ThingSpeak [43] is another open source
IoT application and has an open API to store and retrieve data 4.2.1. Ubiquitous Healthcare Monitoring. Sensor-Clouds can
from device assets or things via LAN or using HTTP over the be used for health monitoring by using a number of easily
Internet. With this platform, location tracking applications, available and most often wearable sensors like accelerometer
sensor logging applications, and social network of device sensors, proximity, ambient light and temperature sensors,
International Journal of Distributed Sensor Networks 9

and so forth to collect patient’s health-related data for track- detect the earthquake and volcano explosion before its
ing sleep activity pattern, blood sugar, body temperature, eruption by continuously monitoring them through the use
and other respiratory conditions [44]. These wearable sensor of several numbers of different sensors like strain, temper-
devices must have support of BWI (Bluetooth’s wireless ature, light, image, sound, acceleration, barometer sensors,
interface), UWB (Ultra wideband), and so forth interface for and so forth through the use of wireless sensor networks
streaming of data and are connected wirelessly to any smart- [48]. Through the Sensor-Cloud infrastructure, the sensor
phone through this interface. These smart phone devices instances engaged in environmental monitoring can be used
pretend to function like a gateway between the remote server in parallel with several other sensor instances, for example,
and sensor through the Internet, maybe GPRS/Wi-Fi, or by the healthcare department to avoid any future casualty, or
other sort of gateways. with crop harvesting application services to avoid the damage
To transform this system into services-based structure, caused by bad weather condition.
web-services-based interfaces are used by smart phone device
to connect to the server [45]. The system prototype should 4.2.3. Telematics. Sensor-Clouds can be used for telematics,
have made to be robust, mobile, and scalable. Robust in the meant to deploy the long distance transmission of our
sense means that it should recover itself from circumstances, computerized or information to a system in continuum. It
which may lack connectivity issues due to power (i.e., bat- enables the smooth communication between system and
tery), failure, or gateway cutoff to patient’s wearable devices devices without any intervention [17].
[46]. Mobile in the sense means that it should be capable
of tracking signals into heterogeneous environments; that is, 4.2.4. Google Health. It is a centralization service of Google
it must catch the signals irrespective of whether the patient that provides personal health information [49] and serves
went outside or still resided into the hospital/building. It as cloud health data storages. Google users are allowed to
should be scalable so that it could be deployed easily for monitor their health records by logging into their accounts at
several users concurrently without affecting the performance collaborated cloud health service providers into the Google
metrics. health system. However, in a recent declaration Google has
Finally, such prototype system should be retargetable and announced the discontinuation of this health service.
extensible in nature. Retargetable refers to the fact that it can
handle various displays with distinct form factors and screen
4.2.5. Microsoft HealthVault. This cloud platform is devel-
resolution. It means that the same health applications can
oped by Microsoft to store and maintain health and fitness-
be displayed to any smartphone display like PDA (personal
related information [50]. HealthVault helps users to store,
digital assistant) or to a bigger console device in a hospital
gather, and share their health relevant information and
where doctors, helpers, or nurses may track the acquired
its data can be acquired from several pharmacies, cloud
data or processed information from distance. The extensi-
providers, health employees, health labs, equipment, and
bility aspect requires that if any newer sensing devices are
from the users itself.
introduced into the system for acquiring the patient’s health-
based information, the system should function efficiently and
conveniently without affecting backend server of the services 4.2.6. Agriculture and Irrigation Control (Field Server Sensors).
[47]. In this platform, context awareness can be achieved Sensor-Cloud can be used in the field of agriculture to
that can direct us to derive a better level of emergency monitor the crop fields in order to upkeep it. For this, a
services to the patient [20]. The information regarding recent field server is developed that comprises of a camera sensors,
operational laboratories, missing doses of pills, number of air sensor, temperature sensor, CO2 concentration sensor,
handicaps, and other situations would be helpful in health soil moisture and temperature sensors, and so forth. These
monitoring. The system should not adhere to any changes sensors continuously upload the field data via Wi-Fi access
made into the operating system or intermediate components point to the field owner to track the health of their crops [51].
of sensing devices and is designed in such a way that it This can also be used for harvesting.
would cause minimal disturbance to services provided to
existing end users of the system [15]. The whole scenario for 4.2.7. Earth Observation. A sensor grid is developed for data
monitoring the healthcare department sensors is shown in gathering from several GPS stations, to process, analyze,
Figure 4. manage, and visualize the GPS data [52]. This GPS data would
In this scenario, several numbers of sensors pick up then be uploaded onto the cloud for efficient monitoring,
the patient data, and these accumulated data are uploaded early warning, and decision-making capability for critical
to a server on cloud. If any noise data is found, they are situations like the volcanic eruptions, earthquakes, tsunamis,
filtered using some filtering mechanism on a server. The cyclones, and so forth to the users all around the world.
doctors/health employees, nurses, and others can then access
the patients’ data on cloud through a web service portal after 4.2.8. Transportation and Vehicular Traffic Applications.
being authenticated/permitted by the patient. Sensor-Cloud can be used to provide an efficient, stable,
equilibrium, and sustainable tracking system. Earlier existing
technologies like GPS navigation can only track the status
4.2.2. Environmental Monitoring for Emergency/Disaster and current location of vehicle. On the other hand, when
Detection. In environmental applications, it is possible to vehicle monitoring is implemented using cloud computing,
10 International Journal of Distributed Sensor Networks

Sensor data
transmitted to
cell phone via Healthcare sensors
Zigbee,
Bluetooth, and so on

Using the several numbers of registered

Infrastructure sensors IT resources in cloud, the data are
processed here to make it meaningful.

Accumulated data sent to

central server via WiFi /GPRS

Server

Data pulled daily by a specific

task to be processed.

Processed results are

pushed back to server
on cloud.

Data recorded manually from

the manual health diary of
Data stored as files arranged in patient to validate the data by
directories means one directory the healthcare staff.
per patient, and also filenames have
timestamps.
Health web portal

Resulting data can then be accessed via web portal for

the health analysis of patient and also the
environmental data are fetched for visualization.

Figure 4: Sensor-Cloud implementation in healthcare.

it is possible to incorporate centralized web service, GPS and (iv) to find the total distance covered,
GSM enabled devices, and embedded device with sensors (v) to track the level of fuel.
[36], which will provide the following benefits:
All the data fetched are stored onto some centralized
(i) to identify the current name of the location, server that will be resided into the cloud. The vehicle owner
(ii) to predict the time of arrival, can access this data on cloud via web portal and can
retrieve all data on cloud in real time to visualize the vehicle
(iii) to find status of driver via alcohol breath sensor, information.
International Journal of Distributed Sensor Networks 11

4.2.9. Tunnel Monitoring. WSN can be used to implement the Web services mainly refers to access the services over Internet
distributed sensing of light levels inside the tunnel and under- connection. It has WSDL (Web Service Description Lan-
bridges to provide necessary input information for adapting guage) definitions, which describe what the web service can
light functionality. This tunnel information can be put onto do, how a web service can be used by client applications,
the cloud and is used to monitor the light intensity in real and where the web service is located. SOAP messages are
time to avoid the automobile users (drivers) casualty and to used to communicate with web services, and these SOAP
save the energy spent unnecessarily for lightening throughout messages are XML based that are transported over the
the day [53]. Internet protocols like SMTP, FTP, and HTTP.

4.2.10. Wildlife Monitoring. Sensor-Cloud can also be used 6. Comparison of Different Approaches in
for tracking the wildlife sanctuaries, forests, and so forth to Sensor-Cloud Infrastructure
regularly monitor the endangered species in real time.
In this section, we first present the advantages and disadvan-
tages of Sensor-Cloud infrastructure in terms of agility, relia-
5. Multiservice Provisioning on bility, portability, real-time, and flexibility. Next, we provide a
Multiple Platforms technical comparison of different messaging approaches and
algorithms used in several existing research on Sensor-Cloud.
Integrating the WSN into heterogeneous networks is a typical Pros of Sensor-Cloud Infrastructure:
task. The reason behind this is the absence of standardized
data exchange functions, which may support the partici- (i) Service requesters or end users can control the service
pating subnetworks of heterogeneous network. Using XML instances freely [12].
in sensor networks encourages the interchangeability of (ii) End users can inspect the status of their relevant
different types of sensors and systems. The lower part of virtual sensors.
Figure 5 deals with the XML [24] encoded physical sensors. (iii) Service requesters can use the virtual sensors without
The XML encoding defines some set of rules for these physical worrying about the implementations detail [12].
sensors such that it will be both human readable and machine
readable with less intervention and will enable these to be (iv) The client/users need not to worry about the exact
implemented on several numbers of different platforms. XML locations and detailed description of their sensors
enables documents to give physical sensor’s metadata, that [45].
is, the type of the physical sensors, its specifications, the (v) The service instances are automatically provisioned
accuracy or intensity of these physical sensors, the exact whenever a request is made [19].
location, and so forth. But sensor nodes have limited storage (vi) The IT resources and sensors are released as and when
and power constraints, and conflict may occur while using the required job is over, which means that users can
the XML encoding. For this reason, the XML support should delete them when they become nonuseful.
be based on efficient data binding techniques to preserve the
(vii) Usage of physical sensors can be tracked by the sensor
time, space, and energy by minimizing the XML overhead
owner.
[34]. This whole scenario can be depicted by Figure 5.
The author in [54] has proposed to access the sensor (viii) Sensor data are available all the time for a number of
information by using the Web Service Description Language various applications until the connection is provided
(WSDL) and structure data, so that the multiple applications [12].
may access sensor information. But the key issue in using (ix) The Sensor-Cloud architecture provides an extensi-
the web services on sensor nodes is energy and bandwidth ble, open, interoperable, and intelligent sensor net-
overhead of structured data formats used in the web services work for service provisioning in health care [56].
[34]. In heterogeneous sensor networks, integration is a
(x) The cost of IT resources and WSN infrastructure is
complex task because there is an absence of standardized data
reduced when integrating with Internet/Cloud [57].
exchange format between the heterogeneous systems and
networks. XML has evolved to overcome this insufficiency (xi) End users can also create the sensors group dynami-
by providing a standard data exchange format between cally in the form of virtual-sensor groups to innovate
heterogeneous network and systems. Because of the limited the new services [12].
hardware resources within sensor networks, XML usage was Besides these advantages, the Sensor-Cloud infrastruc-
not fully introduced earlier. But now XML usage in sensor ture also has some drawbacks and these are as follow.
networks is made applicable by introducing the XML tem- Cons of Sensor-Cloud Infrastructure:
plate objects in an optimized manner [55]. XML is basically a
key feature towards the service-oriented sensor networks and (i) The IT resources and physical sensors should be
a proper medium to support complex data management and prepared prior to operation of the Sensor-Cloud
heterogeneous sensor networks. infrastructure [8].
To enable the applications to communicate with each (ii) The Sensor-Cloud infrastructure will not provide
other and to provide remote access to the services offered much accurate data as in the case of direct sharing of
by Sensor-Cloud platform, web services are introduced [37]. physical sensors data [12].
12 International Journal of Distributed Sensor Networks

User/service register Application 1 OS

Application 2
Application 3 OS

Android apps
··· OS

···
Web services
Application “ 𝑛 ” OS “ 𝑛 ”

SOAP-or
XML-based Sensor cloud infrastructure
messages
Sensor management

XML
encoding and
parsing

···
Physical sensors 1 Physical sensors 2 Physical sensors “𝑛”

Wireless sensor network

Figure 5: XML encoded physical sensors into the Sensor-Cloud.

(iii) Sensor-Cloud infrastructure is vulnerable and more 7. Issues and Challenges While
prone to sophisticated distributed intrusion attacks Designing Sensor-Cloud Infrastructure in
like DDOS (distribute denial of service) and XSS Several Applications
(cross-site scripting) [58].
There are several issues like designing, engineering, reliable
connection, continuous data flow, power issues, and so forth
(iv) A continuous data connectivity is needed between that need to be handled while proposing Sensor-Cloud
end users and Sensor-Cloud server [15]. infrastructure for health care and other different applications
[68]. Some of the main issues are as follows.

Because of continuous processing and wireless transmis- 7.1. Design Issues. There are several issues while designing the
sion, power is another factor that has to be dealt with carefully system in real scenario like nursing home, health care, hos-
in CSI [45]. pitals, and so forth, which require fault-tolerant and reliable
We now highlight the differences among the existing continuous transfer of data from sensor devices to the server.
works on Sensor-Cloud in terms of message passing proto- For example, in a private health care, the patient may be out
cols, data caching, QoS routing, processing, event matching, of coverage area from the smart-phone gateway because of
and other algorithms, as shown in Table 1. patients coming in and out frequently. This scenario would
Table 1: Technical comparison of different messaging approaches and algorithms being used in several papers of Sensor-Cloud.
Using data caching Using Using Using orthogonal Using
Using traditional Using SPWS Using QoS-based Using Using
Papers or hop-by-hop event-matching matrix-matching neural network load-balancing
SOAP messages messages routing SGIM SPMC
processing algorithms algorithms algorithm algorithm
[59] No No No No Yes Yes Yes No No No
International Journal of Distributed Sensor Networks

[5] No No Yes Yes No Yes Yes No No No

[34] No Yes No No No No No No No No
[60] No No Yes No No No Yes No No No
[61] No No No Yes No No Yes No No No
[39, 62] No No No No No No Yes Yes No No
[13, 63] No No Yes No No No No No No No
[64] No No No No No No No No Yes No
[65] No No Yes Yes No No No No No No
[37, 66,
Yes No No No No No No No No No
67]
[45] Yes No No No No No No No No Yes
SPWS: sensor profile for web services.
SGIM: statistical group index matching.
SPMC: stream monitoring and processing component.
13
14 International Journal of Distributed Sensor Networks

be more prone to connection failure between the server and 7.6. Service Level Agreement (SLA) Violation. Consumers
smart phone (or any other display device, like PDA) and thus dependency on cloud providers for their applications’ com-
this scenario must be considered while designing such system puting needs (i.e. their processing, storage, and analysis of
in order to avoid accumulation of errors [15, 69]. enormous sensor data) on demand may require a specific
Quality of Service (QoS) to be maintained. But if cloud
7.2. Storage Issues. Some engineering issues like storage of providers are unable to provide QoS on user’s demand
data at server side and transferring data from phone to server even in the case of processing huge sensor data in critical
must have to be considered. To tackle this, timestamps are environmental situations, it would result in SLA violation
sent with each data packet to assist in reconstruction of data and cloud provider must be responsible for that. So, we
on the server side. Most of the data processing is done at need a reliable dynamic collaboration among cloud providers.
server end so the system must be designed to avoid the bursty But opting for the best combination of cloud providers in
processing due to multiple users connected simultaneously dynamic collaboration is a big challenge in terms of cost, time,
to the system. The system must be designed to accommodate and discrepancy between providers and QoS [38].
multiple users to connect at the same time [15].
Storage issues can be tackled with the introduction of pre- 7.7. Need for Efficient Information Dissemination. In Sensor-
dictive storage concept proposed by the authors in [70]. This Cloud an efficient information dissemination mechanism
concept of storage keeps it easily fit to the correlated behavior is needed that can match the published events or sensor
of the physical environment and builds an architecture that data to appropriate user’s applications. But there are some
focuses the sensor data archival at some remote sensors of issues like maintaining flexibility in providing a powerful
Sensor-Cloud infrastructure. It also uses predictive caching subscription schema, which may capture information about
at proxies. events, guaranteeing the scalability with respect to a number
of subscribers and published events or sensor data [38].
7.3. Authorization Issues. A web-based user interface is Since the data sets and their relevant access services are
used for doctors, patients, helpers, care-givers, and so forth distributed geographically, the allocation of data storage and
to inspect and analyze the patients’ health-related results dissemination becomes critical challenges.
remotely. Therefore, the system should offer different autho-
rization roles for different types of users and authenticated via 7.8. Security and Privacy Support Issues. There are fewer stan-
this web interface. This will enable the privacy to some extent dards available to ensure the integrity of the data in response
by allowing the care givers to restrict them to the patients that to change due to authorized transactions. The consumers
he/she will take care of. need to know whether his/her data at cloud center is well
encrypted or who supervises the encryption/decryption keys
7.4. Power (Battery) Issues. While using smart phone as a (i.e., the cloud vendor or customer himself). Private health
gateway, power (battery) is the main issue that has to be data may become public due to fallacy or inaccuracy; that is,
taken care of because the continuous processing and wireless consumer’s privacy may be lost into cloud and sensor data
transmission would drain out the mobile battery within few or information uploaded into clouds may not be supervised
hours or days. Thus, it is important to tackle power issues correctly by user. The US WellPoint disclosed that 130,000
while connecting mobile phone gateway with the Sensor- records of its consumers had leaked out and become available
Cloud infrastructure [45]. publicly over the Internet. So better privacy policies are the
demand of the time that can offer the services themselves
while maintaining the privacy [37, 38].
7.5. Event Processing and Management. Sensor-Cloud has to
cope with very complex event processing and management
issues [16, 17], such as the following. 7.9. Real-Time Multimedia Content Processing and Massive
Scaling. Usage of large amount of multimedia data and
(i) How the events have to be synchronized that may information in real time and its mining is a big challenge in
come from different sources in different time because the integration of heterogeneous and massive data sources
of delays in network? with cloud. To classify this real-time multimedia information
and contents such that it may trigger the relevant services and
(ii) How the event processing rules have to be changed assist the user in his current location is also a big challenge to
without affecting the system? be handled [16].
(iii) How the messages and events of varying types are
supported? 7.10. Collective Intelligence Harvesting. The heterogeneous
real-time sensor data feed enhances the decision-making
(iv) How to support the enormous numbers of events and
capability by using the appropriate data and decision level
its conditions in an optimal way?
fusion mechanisms. But maximization of intelligence devel-
(v) How can we recognize the context (i.e., spatial, tem- oped from the massively collocated information in cloud is
poral, semantic) to its relevant situation detection? still a very big challenge [16].
International Journal of Distributed Sensor Networks 15

7.11. Energy Efficiency Issues. The basic disadvantages of a different customer’s management, services management, and
WSN and cloud computing are almost the same, and energy modes and methods of payments and pricing. So all this
efficiency of sensor nodes is lost due to the limited storage together will lead to a number of issues [10] such as
and processing capacity of nodes. The authors of [37] have
proposed a system for health monitoring using the textile (i) how to set the price?
sensors, which work much better and give more accurate (ii) how the payment is to be made?
results. These textile sensors can be easily sewed and are even
washable. Although the proposed system of textile sensors (iii) how the price is be distributed among different
is performing well in the majority of aspects, the battery entities?
can last only 24 hours after continuous monitoring and
data transmitting regarding user’s heartbeat rate, movement, The growing demand for controlling and monitoring the
respiratory conditions, and so forth. The gathered accumu- environment and its applications results in the growth of
lated data can then be visualized in charts using some web a large number of devices while the cost of deployment
applications and the results are received at user end through and connecting them to heterogeneous network continues
an alert message remotely on user’s smartphone. But in to drop. However, the interfaces, protocols, connections,
order to extend system independency, energy efficiency of and so forth increase at exponential rate, thereby making
such systems (textile sensors and microcontroller based) is it difficult and expensive for information technology (IT)
a primary issue that has to be handled. people to integrate the devices (sensor devices) into the cloud
Data caching mechanism [71] can be used to reuse bygone world. To eradicate the complexity and cost associated with
sensor data for applications that are tolerant to time, for integrating the sensors into cloud or any highly distributed
example, an application related to variant room temperature. system, the authors of [34] proposed emerging and existing
If this bygone sensor data is used to satisfy the various standards from both domains, for example, embedded sensor
requests for a common sensor data, the energy consumption and IT domains within a service-oriented sensor architecture
will be reduced [11]. Still more work is needed to overcome (SOSA).
the energy consumption. The authors in [34] extend the service oriented paradigm
To improve the energy efficiency and memory usage in to a sensor network and use the service oriented process
a Sensor-Cloud infrastructure, there should be a middleware parameter (like profiling sensors for web services), which
which can tackle the adverse situation in case of continuous helps in intelligence integration into the Internet. This
and long-duration monitoring of data. This can be done solution when extended to Sensor-Cloud would result in
through the gateway that is acting as a middleware and high availability and reliability. It was found that the energy
collects the huge sensor data from sensor nodes [13]. This consumption of sensor nodes reduced drastically when the
middleware should be able to compress the sensor data to data exchange is done among sensors into a heterogeneous
avoid the transmission load and then transmits it back to network with gateway through the SPWS (sensor profiles for
the gateway acting as a middleware on cloud side which in web services) as compared to traditional SOAP messages.
turn decompresses and stores it there. When the transmission It has been observed also that the cost of memory usage
overload reduces, the energy consumption of sensor nodes in sensor nodes remains constant with SPWS whereas it is
improves automatically due to less processing. increased with SOAP messages.
The entity which is most responsible for the cost of
Sensor-Cloud service model is the message exchanged among
7.12. Bandwidth Limitation. Bandwidth limitation is one of sensors into a heterogeneous network environment. Using
the current big challenges that has to be handled in Sensor- SPWS, power consumption of sensor nodes as well as the
Cloud system when the number of sensor devices and their memory usages in sensor nodes are reduced drastically. In
cloud users increases dramatically [71]. However, there are the Sensor-Cloud infrastructure, the cost of communication
a number of optimal and efficient bandwidth allocation among sensors is more than the processing cost. Hence the
methods proposed, but to manage the bandwidth allocation reduction in power consumption and memory usage leads
with such a gigantic infrastructure consisting of huge device to less communication cost which also enables an energy
assets and cloud users, the task of allocating bandwidth to efficient model of Sensor-Cloud.
every devices and users becomes very difficult.
7.15. Interface Standardization Issues. Web interfaces cur-
7.13. Network Access Management. There are various num- rently provide the interface among Sensor-Cloud users (may
bers of networks to deal with in Sensor-Cloud architecture be smartphone users) and cloud. But web interface may
applications. So a proper and efficient access management cause overhead because the web interfaces are not specif-
scheme for these networks is needed because this will ically designed for smart phones or mobile devices. Also,
optimize the bandwidth usage and improve link performance there would be compatibility issues for web interface among
[72]. devices and in this case signaling, standard protocol, and
interface for interacting between Sensor-Cloud users and the
7.14. Pricing Issues. Access to the services of Sensor-Cloud cloud would require seamless services for implementation.
involves both the sensor-service provider (SSP) and cloud- Thus, interoperability would be a big issue when the Sensor-
service provider (CSP). However, both SSPs and CSPs have Cloud users need to access the services with cloud [10].
16 International Journal of Distributed Sensor Networks

7.16. Maintenance Issues. In order to keep the end users’ [4] G. Simon, G. Balogh, G. Pap et al., “Sensor network-based
loyalty, the cloud should cope with the service failure [73]. countersniper system,” in Proceedings of the 2nd International
For this a regular maintenance is needed and redundancy Conference on Embedded Networked Sensor Systems (SenSys
techniques should be implemented to ensure the smooth and ’04), pp. 1–12, Baltimore, Md, USA, November 2004.
continuous flow of services. This can be done by backing [5] S. K. Dash, J. P. Sahoo, S. Mohapatra, and S. P. Pati, “Sensor-
up the data regularly and by distributing their multiple data cloud: assimilation of wireless sensor network and the cloud,”
centers geographically across the world. in Advances in Computer Science and Information Technology.
Networks and Communications, vol. 84, pp. 455–464, Springer-
Link, 2012.
7.17. Resource and Hardware Compatibility Issues. Hardware [6] M. Castillo-Effen, D. H. Quintela, R. Jordan, W. Westhoff, and
compatibility as well as software compatibility both can be W. Moreno, “Wireless sensor networks for flash-flood alerting,”
solved in cloud computing environment by allowing the in Proceedings of the 5th IEEE International Caracas Conference
sharing of hardware or software resources or services [5], but on Devices, Circuits and Systems (ICCDCS ’04), pp. 142–146,
there may be the case when sensor or some other resources November 2004.
being used are lost due to some calamity or severe weather [7] G. Werner-Allen, K. Lorincz, M. Welsh et al., “Deploying a
condition. To handle these issues, the authors in [70] have wireless sensor network on an active volcano,” IEEE Internet
proposed PRESTO architecture that enables users/clients to Computing, vol. 10, no. 2, pp. 18–25, 2006.
view the data in a single logical view which is distributed [8] M. Yuriyama, T. Kushida, and M. Itakura, “A new model of
across several sensor proxies and remote sensors. This enables accelerating service innovation with sensor-cloud infrastruc-
the users to view variability at several lossy levels and ture,” in Proceedings of the annual SRII Global Conference (SRII
unreliable remote sensor network resources. The PRESTO ’11), pp. 308–314, 2011.
also supports archival queries on data resources that enable [9] J. Yick, B. Mukherjee, and D. Ghosal, Wireless Sensor Network
the historical data query to prevent any unusual events to Survey, Elsevier, 2008.
better understand them for future application scenarios. [10] H. T. Dinh, C. Lee, D. Niyato, and P. Wang, A Survey of Mobile
Cloud Computing: Architecture, Applications, and Approaches,
Wireless Communications and Mobile Computing-Wiley
8. Conclusion Online Library, 2011.
[11] W. Kim, “Cloud computing: today and tomorrow,” Journal of
In this paper, we surveyed the use of Sensor-Cloud architec- Object Technology, vol. 8, pp. 65–72, 2009.
ture in the context of several applications. The Sensor-Cloud
[12] M. Yuriyama and T. Kushida, “Sensor-cloud infrastructure
architecture enables the sensor data to be categorized, stored, physical sensor management with virtualized sensors on cloud
and processed in such a way that it becomes cost-effective, computing,” in Proceedings of the IEEE 13th International
timely available, and easily accessible. Earlier, most WSN sys- Conference on Network-Based Information Systems (NBiS ’10),
tems which were included to several controlling/monitoring pp. 1–8, September 2010.
schemes were closed in nature, zero, or less interoperability, [13] L. P. D. Kumar, S. S. Grace, A. Krishnan, V. M. Manikandan,
specific application oriented, and nonextensible. However, R. Chinraj, and M. R. Sumalatha, “Data filtering in wireless
integrating the existing sensors with cloud will enable an sensor networks using neural networks for storage in cloud,”
open, extensible, scalable, interoperable, and easy to use, re- in Proceedings of the IEEE International Conference on Recent
constructible network of sensors for numerous applications. Trends in Information Technology (ICRTIT ’11), 2012.
In this paper, we have discussed the opportunities of imple- [14] M. O’Brien, Remote Telemonitoring—A Preliminary Review of
menting the technology to handle more complex situations Current Evidence, European Center for Connected Health,
of a real world through the service innovation capability of 2008.
Sensor-Cloud infrastructure. [15] B. Jit, J. Maniyeri, K. Gopalakrishnan et al., “Processing of
wearable sensor data on the cloud—a step towards scaling of
continuous monitoring of health and well-being,” in Proceedings
Acknowledgment of the 32nd Annual International Conference of the IEEE Engi-
neering in Medicine and Biology Society (EMBC ’10), pp. 3860–
The authors extend their appreciation to the Deanship of 3863, 2010.
Scientific Research at King Saud University for funding this [16] http://www.ntu.edu.sg/intellisys.
work through the research group project no. RGP-VPP-049.
[17] K. T. Lan, “What’s Next? Sensor+Cloud?” in Proceeding of the
7th International Workshop on Data Management for Sensor
References Networks, pp. 978–971, ACM Digital Library, 2010.
[18] Sensor-Cloud, http://sensorcloud.com/system-overview.
[1] K. Romer and F. Mattern, “The design space of wireless sensor [19] C. O. Rolim, F. L. Koch, C. B. Westphall, J. Werner, A. Fracalossi,
networks,” IEEE Wireless Communications, vol. 11, no. 6, pp. 54– and G. S. Salvador, “A cloud computing solution for patient’s
61, 2004. data collection in health care institutions,” in Proceedings of
[2] T. Haenselmann, “Sensor networks,” GFDL Wireless Sensor the 2nd International Conference on eHealth, Telemedicine, and
Network textbook 2006. Social Medicine (eTELEMED ’10), pp. 95–99, February 2010.
[3] I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, “A [20] U. Varshney, “Pervasive healthcare and wireless health moni-
survey on sensor networks,” IEEE Communications Magazine, toring,” Mobile Networks and Applications, vol. 12, no. 2-3, pp.
vol. 40, no. 8, pp. 102–114, 2002. 113–127, 2007.
International Journal of Distributed Sensor Networks 17

[21] LifeShirt, http://www.vivometrics.com/site/system.html. [38] X. H. Le, “Secured WSN-integrated cloud computing for u-
[22] Smart Shirt, http://www.gtwm.gatech.edu. life care,” in Proceedings of the Consumer Communications and
networking Conference (CCNC ’10), pp. 1–2, IEEE, 2010.
[23] U. Varshney, “Managing wireless health monitoring for people
with disabilities,” IEEE IT-Professional, pp. 12–16, 2006. [39] T.-D. Nguyen and E.-N. Huh, “An efficient key management for
secure multicast in Sensor-Cloud”,” in Proceedings of the IEEE
[24] SensorML, http://www.opengeospatial.org/standards/sensor-
1st ACIS/JNU International Conference on Computers, Networks,
ml.
Systems, and Industrial Engineering, 2011.
[25] O. Gnawali, R. Fonseca, K. Jamieson, D. Moss, and P. Levis,
[40] Nimbits Data Logging Cloud Sever, http://www.nimbits.com.
“Collection tree protocol,” in Proceedings of the 7th ACM
Conference on Embedded Networked Sensor Systems (SenSys [41] Pachube Feed Cloud Service, http://www.pachube.com.
’09), pp. 1–14, November 2009. [42] iDigi—Device Cloud, http://www.idigi.com.
[26] A. Rowe, V. Gupta, and R. Rajkumar, “Low-power clock [43] IoT—ThingSpeak, http://www.thingspeak.com.
synchronization using electromagnetic energy radiating from [44] G. Demiris, B. K. Hensel, M. Skubic, and M. Rantz, “Senior
AC power lines,” in Proceedings of the 7th ACM Conference on residents’ perceived need of and preferences for “smart home”
Embedded Networked Sensor Systems (SenSys ’09), pp. 211–224, sensor technologies,” International Journal of Technology Assess-
November 2009. ment in Health Care, vol. 24, no. 1, pp. 120–124, 2008.
[27] S. Madden and M. J. Franklin, “Fjording the stream: an archi- [45] K. Lee, D. Murray, D. Hughes, and W. Joosen, “Extending sensor
tecture for queries over streaming sensor data,” in Proceedings of networks into the Cloud using Amazon web services,” in Pro-
the 18th International Conference on Data Engineering, pp. 555– ceedings of the 1st IEEE International Conference on Networked
566, March 2002. Embedded Systems for Enterprise Applications (NESEA ’10), pp.
[28] R. Katsuma, Y. Murata, N. Shibata, K. Yasumoto, and M. Ito, 1–7, November 2010.
“Extending k-coverage lifetime of wireless sensor networks [46] B. Jit, J. Maniyeri, S. Louis, K. Gopalakrishnan, and P. Yap,
using mobile sensor nodes,” in Proceedings of the 5th IEEE “Design and trial deployment of a practical sleep activity
International Conference on Wireless and Mobile Computing pattern monitoring system,” in Proceedings of the International
Networking and Communication (WiMob ’09), pp. 48–54, Octo- Conference on Smart Homes and Health Telematics (ICOST ’09),
ber 2009. Tours, France, June 2009.
[29] J. Hill, R. Szewczyk, A. Woo, S. Hollar, D. Culler, and K. [47] B. Jit, J. Maniyeri, S. Louis, and L. K. P. Yap, “Fast matching
Pister, “System architecture directions for networked sensors,” of sensor data with manual observations,” in Proceedings of the
in Proceedings of the 9th Internatinal Conference Architectural 31st Annual International Conference of the IEEE Engineering
Support for Programming Languages and Operating Systems, pp. in Medicine and Biology Society: Engineering the Future of
93–104, November 2000. Biomedicine (EMBC ’09), pp. 1675–1678, September 2009.
[30] K. Matsumoto, R. Katsuma, N. Shibata, K. Yasumoto, and M. Ito, [48] N. Kurata, M. Suzuki, S. Saruwatari, and H. Morikawa, “Actual
“Extended abstract: minimizing localization cost with mobile application of ubiquitous structural monitoring system using
anchor in underwater sensor networks,” in Proceedings of the wireless sensor networks,” in Proceedings of the 14th World
4th ACM International Workshop on UnderWater Networks Conference on Earthquake Engineering (WCEE ’08), 2008.
(WUWNet ’09), November 2009. [49] Google Health, http://www.google.com/health.
[31] T. Sookoor, T. Hnat, P. Hooimeijer, W. Weimer, and K. White- [50] Korea u-Life care system.
house, “Macrodebugging: global views of distributed program [51] http://www.apan.net/meetings/HongKong2011/Session/Agri-
execution,” in Proceedings of the 7th ACM Conference on culture.php/.
Embedded Networked Sensor Systems (SenSys ’09), pp. 141–154,
November 2009. [52] H. H. Tran and K. J. Wong, “Mesh networking for seismic
monitoring—the sumatran cGPS array case study,” in Pro-
[32] J. Shneidman, P. Pietzuch, J. Ledlie, M. Roussopoulos, and M. ceedings of the IEEE Wireless Communications and Networking
Seltzerand Welsh, “Hourglass: an infrastructure for connecting Conference (WCNC ’09), April 2009.
sensor networks and applications,” Harvard Technical Report
TR-21-04, 2004. [53] Tunnel Monitoring System: http://www.advantech.com/int-
elligent-automation/Industry%20Focus/%7BC274D52C-95-
[33] M. Gaynor, S. L. Moulton, M. Welsh, E. LaCombe, A. Rowan, D2-499E-9E16-6C1F41D1CD6/.
and J. Wynne, “Integrating wireless sensor networks with the
[54] N. B. Priyantha, A. Kansal, M. Goraczko, and F. Zhao, “Tiny
grid,” IEEE Internet Computing, vol. 8, no. 4, pp. 32–39, 2004.
web services: design and implementation of interoperable and
[34] R. S. Ponmagal and J. Raja, “An extensible cloud architecture evolvable sensor networks,” in Proceedings of the 6th ACM
model for heterogeneous sensor services,” International Journal Conference on Embedded Networked Sensor Systems (SenSys
of Computer Science and Information Security, vol. 9, no. 1, 2011. ’08), pp. 253–266, Raleigh, NC, USA, 2008.
[35] S. Guo, Z. Zhong, and T. He, “FIND: faulty node detection [55] N. Hoeller, C. Reinke, J. Neumann, S. Groppe, D. Boeckmann,
for wireless sensor networks,” in Proceedings of the 7th ACM and V. Linnemann, “Efficient XML usage within wireless sensor
Conference on Embedded Networked Sensor Systems (SenSys networks,” in Proceedings of the 4th Annual International Con-
’09), pp. 253–266, November 2009. ference on Wireless Internet (WICON ’08), pp. 17–19, November
[36] A. Alexe and R. Exhilarasie, “Cloud computing based vehicle 2008.
tracking information systems,” International Journal of Com- [56] A. Triantafyllidis, V. Koutkias, I. Chouvarda, and N. Maglav-
puter Science and Telecommunications, vol. 2, no. 1, 2011. eras, “An open and reconfigurable Wireless Sensor Network
[37] C. Doukas and I. Maglogiannis, “Managing wearable sensor for pervasive health monitoring,” in Proceedings of the 2nd
data through cloud computing,” in Proceedings of the IEEE 3rd International Conference on Pervasive Computing Technologies
International Conference on Cloud Computing, 2011. for Healthcare, pp. 112–115, February 2008.
18 International Journal of Distributed Sensor Networks

[57] K. Lee and D. Hughes, “System architecture directions for [71] Y. Xu, S. Helal, T. My Thai, and M. Schmalz, “Optimiz-
tangible cloud computing,” in Proceedings of the International ing push/pull envelopes for energy-efficient cloud-sensor sys-
Workshop on Information Security and Applications (IWISA ’10), tems,” in Proceedings of the 14th ACM international conference
Qinhuangdao, China, October 2010. (MSWiM ’11), 2011.
[58] I. Gul and M. Hussain, “Distributed cloud intrusion detection [72] F. Ge, H. Lin, A. Khajeh et al., “Cognitive radio rides on the
model,” International Journal of Advanced Science and Engineer- cloud,” in Proceedings of the IEEE Military Communications
ing Technology, vol. 34, 2011. Conference (MILCOM ’10), pp. 1448–1453, November 2010.
[59] M. M. Hassan, B. Song, and E. N. Huh, “A framework of [73] R. Liu and I. J. Wassell, “Opportunities and challenges of
sensor—cloud integration opportunities and challenges,” in wireless sensor networks using cloud services,” in Proceedings of
Proceedings of the 3rd International Conference on Ubiquitous the Workshop on Internet of Things and Service Platforms (IoTSP
Information Management and Communication (ICUIMC ’09), ’11), Tokyo, Japan, December 2011.
pp. 618–626, ACM, Suwon, Republic of Korea, January 2009.
[60] J. Heidemann, F. Silva, C. Intanagonwiwat, R. Govindan, D.
Estrin, and D. Ganesan, “Building efficient wireless sensornet-
works with low-level naming,” in Proceedings of the Symposium
on Operating Systems Principles, pp. 146–159, ACM, Alberta,
Canada, October 2001.
[61] S. K. Dash, S. Mohapatra, and P. K. Pattnaik, “A survey on
applications of Wireless Sensor Network using Cloud Comput-
ing,” International Journal of Computer Science and Emerging
Technologies, vol. 1, no. 4, 2010.
[62] K. Ahmed and M. Gregory, “Integrating wireless sensor net-
works with cloud computing,” in Proceedings of the 7th Interna-
tional Conference on Mobile Ad-hoc and Sensor Networks (MSN
’11), pp. 364–366, Beijing, China, 2011.
[63] R. Ian, C. Michele, J. Ho Young, M. Zoltan, and A. Karl,
“Building a front end interface for sensor data cloud,” in
Proceedings of the 2nd International Conference on Signals,
Systems and Automation (ICSSA ’11), vol. 6784 of Lecture Note
in Computer Science, 2011.
[64] J. Cen, T. Yu, Z. Li, S. Jin, and S. Liu, “Developing a disaster
surveillance system based on wireless sensor network and cloud
platform,” in Proceedings of the IET International Conference
on Communication Technology and Application (ICCTA ’11), pp.
757–761, Beijing, China, 2012.
[65] M. Baktashmotlagh, A. Bigdeli, and B. C. Lovell, “Dynamic
resource aware sensor networks: Integration of sensor cloud and
ERPs,” in Proceedings of the 8th IEEE International Conference
on Advanced Video and Signal Based Surveillance (AVSS ’11), pp.
455–460, Klagenfurt, Austria, 2011.
[66] V. V. Tan, D. S. Yoo, and M. J. Yi, “Design and implementation
of Web service by using OPC XML-DA and OPC complex data
for automation and control systems,” in Proceedings of the 6th
IEEE International Conference on Computer and Information
Technology (CIT ’06), p. 263, IEEE, September 2006.
[67] V. Rajesh, J. M. Gnanasekar, R. S. Ponmagal, and P. Anbal-
agan, “Integration of wireless sensor network with cloud,” in
Proceedings of the International Conference on Recent Trends in
Information, Telecommunication, and Computing (ITC ’10), pp.
321–323, March 2010.
[68] C. O. Rolim, F. L. Koch, A. Sekkaki, and C. B. Westphall,
“Telemedicine with grids and wireless sensors networks,” in
Proceedings of the International Conference on e-Medical Systems
(e-Medisys ’08), 2008.
[69] G. Singh, J. O’Donoghue, and C. K. Soon, “Telemedicine: issues
and Implications,” Technology and Health Care, vol. 10, no. 1, pp.
1–10, 2002.
[70] M. M. Islam, M. M. Hassan, G. W. Lee, and E. N. Huh, “A survey
on virtualization of wireless sensor networks,” Sensors Journal,
vol. 12, no. 2, pp. 2175–2207, 2012.