AnIntegratedCloud-basedFrameworkforMobilePhoneSensingRasool FakoorDept. of Comp. Sci. and Eng. Computer Science Engineering (CSE) Notes | EduRev

Computer Science Engineering (CSE) : AnIntegratedCloud-basedFrameworkforMobilePhoneSensingRasool FakoorDept. of Comp. Sci. and Eng. Computer Science Engineering (CSE) Notes | EduRev

 Page 1


AnIntegratedCloud-basedFramework
forMobilePhoneSensing
Rasool Fakoor
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
rasool.fakoor@mavs.uta.edu
Mayank Raj
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
mayank.raj@mavs.uta.edu
Azade Nazi
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
azade.nazi@mavs.uta.edu
Mario Di Francesco
Dept. of Comp. Sci. and Eng.
Aalto University
Espoo, Finland
mario.di.francesco@aalto.?
Sajal K. Das
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
das@uta.edu
ABSTRACT
Nowadays mobile phones are not only communication de-
vices, but also a source of rich sensory data that can be
collected and exploited by distributed people-centric sensing
applications. Among them, environmental monitoring and
emergency response systems can particularly bene?t from
people-based sensing. Due to the limited resources of mo-
bile devices, sensed data are usually o?oaded to the cloud.
However, state-of-the art solutions lack a uni?ed approach
suitable to support diverse applications, while reducing the
energy consumption of the mobile device. In this paper, we
speci?cally address mobile devices as rich sources of multi-
modal data collected by users. In this context, we propose
an integrated framework for storing, processing and deliv-
ering sensed data to people-centric applications deployed in
the cloud. Our integrated platform is the foundation of a
new delivery model, namely, Mobile Application as a Ser-
vice (MAaaS), which allows the creation of people-centric
applications across di?erent domains, including participa-
tory sensing and mobile social networks. We speci?cally
address a case study represented by an emergency response
system for ?re detection and alerting. Through a prototype
testbed implementation, we show that the proposed frame-
work can reduce the energy consumption of mobile devices,
while satisfying the application requirements.
Categories and Subject Descriptors
C.2.4 [COMPUTER-COMMUNICATION
NETWORKS]: Distributed Systems—Distributed applica-
tions;C.2.1[COMPUTER-COMMUNICATION
NETWORKS]: Network Architecture and Design —Dis-
tributed networks
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for pro?t or commercial advantage and that copies
bear this notice and the full citation on the ?rst page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior speci?c
permission and/or a fee.
MCC’12, August 17, 2012, Helsinki, Finland.
Copyright 2012 ACM 978-1-4503-1519-7/12/08 ...$15.00.
General Terms
Design, Reliability
Keywords
Mobile Cloud Computing, People-centric Applications, Mo-
bile Application as a Service (MAaaS), Publish/Subscribe,
Mobile Phone Sensing
1. INTRODUCTION
The proliferation of mobile devices, combined with their
diverse sensing capabilities, has made them a ubiquitous
and rich source of sensory data. Such data include loca-
tion (e.g., GPS), orientation (e.g., gyroscope, accelerome-
ter), textual (social) and multimedia (e.g., image, audio and
video) data [9]. They can be jointly used to develop a wide
range of applications aimed at improving the quality of life.
These applications include environmental [12, 3] and tra?c
monitoring [18, 17], navigation [13] and emergency response
systems [10], as well as smart healthcare and well-being [9].
Many of the applications mentioned above are people-
centric, inthesensethattheyassistindividualusers, agroup
of people, or even a large community, by leveraging sen-
sory data from a large number of sources [2]. Depending on
user demands, sensing mechanisms can be either passive or
active, namely, participatory. Even though some of these
applications have been deployed in many environments [2,
9], integrating them into a single platform is still an open
research problem. This is due to the presence of a large
number of applications with contrasting requirements, di-
verse sensing sources and platforms, as well as multi-modal
data.
In this paper, we propose an integrated framework which
allows uni?ed sensing for diverse people-centric applications.
To ensure scalability, availability and robustness of the sens-
ing and application environments, the framework is de?ned
as part of the cloud infrastructure itself. Speci?cally, our in-
tegrated platform is the foundation of a new delivery model
– namely, Mobile Application as a Service (MAaaS) – which
allows the creation of people-centric applications across dif-
ferent domains, including participatory sensing and mobile
social networks. Each mobile device acts as an indepen-
dent multi-modal data source, and the proposed cloud-based
47
Page 2


AnIntegratedCloud-basedFramework
forMobilePhoneSensing
Rasool Fakoor
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
rasool.fakoor@mavs.uta.edu
Mayank Raj
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
mayank.raj@mavs.uta.edu
Azade Nazi
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
azade.nazi@mavs.uta.edu
Mario Di Francesco
Dept. of Comp. Sci. and Eng.
Aalto University
Espoo, Finland
mario.di.francesco@aalto.?
Sajal K. Das
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
das@uta.edu
ABSTRACT
Nowadays mobile phones are not only communication de-
vices, but also a source of rich sensory data that can be
collected and exploited by distributed people-centric sensing
applications. Among them, environmental monitoring and
emergency response systems can particularly bene?t from
people-based sensing. Due to the limited resources of mo-
bile devices, sensed data are usually o?oaded to the cloud.
However, state-of-the art solutions lack a uni?ed approach
suitable to support diverse applications, while reducing the
energy consumption of the mobile device. In this paper, we
speci?cally address mobile devices as rich sources of multi-
modal data collected by users. In this context, we propose
an integrated framework for storing, processing and deliv-
ering sensed data to people-centric applications deployed in
the cloud. Our integrated platform is the foundation of a
new delivery model, namely, Mobile Application as a Ser-
vice (MAaaS), which allows the creation of people-centric
applications across di?erent domains, including participa-
tory sensing and mobile social networks. We speci?cally
address a case study represented by an emergency response
system for ?re detection and alerting. Through a prototype
testbed implementation, we show that the proposed frame-
work can reduce the energy consumption of mobile devices,
while satisfying the application requirements.
Categories and Subject Descriptors
C.2.4 [COMPUTER-COMMUNICATION
NETWORKS]: Distributed Systems—Distributed applica-
tions;C.2.1[COMPUTER-COMMUNICATION
NETWORKS]: Network Architecture and Design —Dis-
tributed networks
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for pro?t or commercial advantage and that copies
bear this notice and the full citation on the ?rst page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior speci?c
permission and/or a fee.
MCC’12, August 17, 2012, Helsinki, Finland.
Copyright 2012 ACM 978-1-4503-1519-7/12/08 ...$15.00.
General Terms
Design, Reliability
Keywords
Mobile Cloud Computing, People-centric Applications, Mo-
bile Application as a Service (MAaaS), Publish/Subscribe,
Mobile Phone Sensing
1. INTRODUCTION
The proliferation of mobile devices, combined with their
diverse sensing capabilities, has made them a ubiquitous
and rich source of sensory data. Such data include loca-
tion (e.g., GPS), orientation (e.g., gyroscope, accelerome-
ter), textual (social) and multimedia (e.g., image, audio and
video) data [9]. They can be jointly used to develop a wide
range of applications aimed at improving the quality of life.
These applications include environmental [12, 3] and tra?c
monitoring [18, 17], navigation [13] and emergency response
systems [10], as well as smart healthcare and well-being [9].
Many of the applications mentioned above are people-
centric, inthesensethattheyassistindividualusers, agroup
of people, or even a large community, by leveraging sen-
sory data from a large number of sources [2]. Depending on
user demands, sensing mechanisms can be either passive or
active, namely, participatory. Even though some of these
applications have been deployed in many environments [2,
9], integrating them into a single platform is still an open
research problem. This is due to the presence of a large
number of applications with contrasting requirements, di-
verse sensing sources and platforms, as well as multi-modal
data.
In this paper, we propose an integrated framework which
allows uni?ed sensing for diverse people-centric applications.
To ensure scalability, availability and robustness of the sens-
ing and application environments, the framework is de?ned
as part of the cloud infrastructure itself. Speci?cally, our in-
tegrated platform is the foundation of a new delivery model
– namely, Mobile Application as a Service (MAaaS) – which
allows the creation of people-centric applications across dif-
ferent domains, including participatory sensing and mobile
social networks. Each mobile device acts as an indepen-
dent multi-modal data source, and the proposed cloud-based
47 48
Page 3


AnIntegratedCloud-basedFramework
forMobilePhoneSensing
Rasool Fakoor
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
rasool.fakoor@mavs.uta.edu
Mayank Raj
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
mayank.raj@mavs.uta.edu
Azade Nazi
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
azade.nazi@mavs.uta.edu
Mario Di Francesco
Dept. of Comp. Sci. and Eng.
Aalto University
Espoo, Finland
mario.di.francesco@aalto.?
Sajal K. Das
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
das@uta.edu
ABSTRACT
Nowadays mobile phones are not only communication de-
vices, but also a source of rich sensory data that can be
collected and exploited by distributed people-centric sensing
applications. Among them, environmental monitoring and
emergency response systems can particularly bene?t from
people-based sensing. Due to the limited resources of mo-
bile devices, sensed data are usually o?oaded to the cloud.
However, state-of-the art solutions lack a uni?ed approach
suitable to support diverse applications, while reducing the
energy consumption of the mobile device. In this paper, we
speci?cally address mobile devices as rich sources of multi-
modal data collected by users. In this context, we propose
an integrated framework for storing, processing and deliv-
ering sensed data to people-centric applications deployed in
the cloud. Our integrated platform is the foundation of a
new delivery model, namely, Mobile Application as a Ser-
vice (MAaaS), which allows the creation of people-centric
applications across di?erent domains, including participa-
tory sensing and mobile social networks. We speci?cally
address a case study represented by an emergency response
system for ?re detection and alerting. Through a prototype
testbed implementation, we show that the proposed frame-
work can reduce the energy consumption of mobile devices,
while satisfying the application requirements.
Categories and Subject Descriptors
C.2.4 [COMPUTER-COMMUNICATION
NETWORKS]: Distributed Systems—Distributed applica-
tions;C.2.1[COMPUTER-COMMUNICATION
NETWORKS]: Network Architecture and Design —Dis-
tributed networks
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for pro?t or commercial advantage and that copies
bear this notice and the full citation on the ?rst page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior speci?c
permission and/or a fee.
MCC’12, August 17, 2012, Helsinki, Finland.
Copyright 2012 ACM 978-1-4503-1519-7/12/08 ...$15.00.
General Terms
Design, Reliability
Keywords
Mobile Cloud Computing, People-centric Applications, Mo-
bile Application as a Service (MAaaS), Publish/Subscribe,
Mobile Phone Sensing
1. INTRODUCTION
The proliferation of mobile devices, combined with their
diverse sensing capabilities, has made them a ubiquitous
and rich source of sensory data. Such data include loca-
tion (e.g., GPS), orientation (e.g., gyroscope, accelerome-
ter), textual (social) and multimedia (e.g., image, audio and
video) data [9]. They can be jointly used to develop a wide
range of applications aimed at improving the quality of life.
These applications include environmental [12, 3] and tra?c
monitoring [18, 17], navigation [13] and emergency response
systems [10], as well as smart healthcare and well-being [9].
Many of the applications mentioned above are people-
centric, inthesensethattheyassistindividualusers, agroup
of people, or even a large community, by leveraging sen-
sory data from a large number of sources [2]. Depending on
user demands, sensing mechanisms can be either passive or
active, namely, participatory. Even though some of these
applications have been deployed in many environments [2,
9], integrating them into a single platform is still an open
research problem. This is due to the presence of a large
number of applications with contrasting requirements, di-
verse sensing sources and platforms, as well as multi-modal
data.
In this paper, we propose an integrated framework which
allows uni?ed sensing for diverse people-centric applications.
To ensure scalability, availability and robustness of the sens-
ing and application environments, the framework is de?ned
as part of the cloud infrastructure itself. Speci?cally, our in-
tegrated platform is the foundation of a new delivery model
– namely, Mobile Application as a Service (MAaaS) – which
allows the creation of people-centric applications across dif-
ferent domains, including participatory sensing and mobile
social networks. Each mobile device acts as an indepen-
dent multi-modal data source, and the proposed cloud-based
47 48
tionshosted on the cloud and select those they are interested
in directly from the MCA.
The rest of the system components is part of the cloud
infrastructure. The User Adaptation Module (UAM) in-
terfaces the MCA with the Mobile Cloud Engine (MCE).
Speci?cally, there is one UAM for each MCA. When mul-
tiple MCEs share a common user (namely, a MCA), the
related device sends the data once to the UAM which, in
turn, sends a copy to the relevant MCEs.
The Storage Module (SM) allows to save and retrieve sen-
sorydata. Itsupportsmulti-modaldata, rangingfromscalar
data (e.g., location, accelerometer and gyroscope readings)
to generic textual information (e.g., summary descriptions
uploaded by the users) and multimedia data (i.e., images,
audio and videos captured through the mobile device). As
a consequence, the SM is data-independent, but provides
means to interconnect to the cloud for mining and knowl-
edge extraction purposes. The major features of the SM are
scalability and responsiveness.
The Application Interface Module (AIM) interacts with
the framework to get the required data. It further allows
the application developer to access computing and storage
resources on-demand for ?exibility and scalability.
Finally, the Mobile Cloud Engine (MCE) processes and
controls the data ?ow in the framework, and is instantiated
for each application. Since the MCE is the core component
of our solution, it will be detailed in a separate subsection
below.
3.1 Mobile Cloud Engine
Our proposed framework contains two types of mobile
cloud engines. The ?rst type is called control MCE or
(MCE0). There is only one MCE0 in the whole framework,
independent of the actual people-centric applications. The
MCE0 is responsible to initiate other MCEs when a new ap-
plication request arrives in the framework. In addition, it
always collects the sensory data from mobile users in pre-
de?ned periods, regardless of the application requests. The
data collected by the MCE0 can be used for di?erent major
purposes, namely: mining; processing; and as a baseline to
start other MCEs. The second class of MCEs is instanti-
ated based on the actual application requests. These MCEs
are mainly employed to deal with speci?c people-centric ap-
plications. However, both classes of MCEs have the same
structure to provide coherency in the framework.
The MCE is composed of three sub-modules. The pub-
lish/subscribe (PS) module is primarily responsible for es-
tablishing the data ?ow between UAMs and the MCEs. The
PS module is composed of control and data message queues.
The control message queue handles all the control messages,
such as the data sending rate. The data message queues are
composed of multiple queues, each associated with a di?er-
ent priority (e.g., low, medium, and high). The data queue
delivers the captured data from the UAM to the associated
applications, and also forwards them to the storage module.
Furthermore, the PS module delivers the processed infor-
mation back to the mobile users. In this framework, the PS
modules are classi?ed in three categories, according to the
related priorities, namely, low, medium, and high. Di?erent
priorities are mainly used to regulate the sampling rate at
the mobile device, the data reporting rate to the cloud, and
the actual data processing within the cloud.
Thedecisionmodule (DM)handlesandregulatesthetrans-
actions among the di?erent parts of the MCE. It also classi-
?es the PS modules based on the requests received from ap-
plications. Speci?cally, once an application request arrives
in the MCE, the decision module identi?es which type of
PS module is needed to satisfy the request. Furthermore, it
recognizes the potential group of mobile users who can par-
ticipate to the application query. Finally, the DM module
initiates the context-awareness module (described next) to
provide people-centric and location-based services. The de-
cision module provides functions to process and mine data.
It can use unsupervised learning approaches to deal with
unstructured data in the system. One example is to ?nd
a coherent group of mobile phone users who share common
interests (e.g., through their frequently visited places) or to
predict user behavior.
Finally, the context-awareness (CA) module provides con-
text information to the di?erent people-centric applications.
In its simplest form, the context is represented by location.
For example, a pollution monitoring application may con-
sider the pollutant concentration at the locations in the
vicinity of the user. In the more general case, context may
include the activity of the user or its social interactions with
other users.
4. CASE STUDY
Texas is very likely to experience wild?res in Summer.
The state of Texas and the federal government spend sub-
stantial amount of money to handle wild?res, in order to en-
sure public safety and reduce damages. For instance, from
November 2010 to October 2011, about 30,457 ?res spread
over 3,993,716 acres and destroyed 3,017 homes [16]. The
cost of damages to houses was estimated to be more than
100 million dollars [4]. Dispatching ?re ?ghters to the exact
location of wild?res and preventing the ?re from spreading
play an important role in e?ciently handling the emergency.
However, placingcamerasandsensorsatalllocationstocon-
tinuously collect information is not feasible due to the high
deployment and maintenance costs.
In this section, we demonstrate how our proposed archi-
tecture can help in this scenario. Let us assume that the
mobile cloud application for emergency response has already
been installed on the user mobile devices and that a ?re hap-
pens in an certain area. In the proposed framework, user
location is periodically updated and stored in the the mo-
bile cloud framework either explicitly by the mobile device
user or through a request from the wild?re monitoring and
response agency. When a wild?re is reported, the agency
initiates a request to the framework with the information of
the a?ected area. After receiving the request, the framework
?nds the people in that area, based on their GPS locations.
Then, the framework sends an emergency noti?cation and
requests for updates on the situation.
At the same time, the mobile cloud framework updates
the agency with current available data in the system un-
til new data arrives. The application can provide di?erent
visualizations, ranging from a map showing the location of
people who are in the vicinity to a live video stream of a cer-
tain place. By using this method, the agency can obtain an
accurate characterization of the wild?re without deploying
any camera or sensor. The application can provide indirect
communication between the agency and mobile phone users
to explicitly obtain additional information which, in turn,
can help increase the accuracy and reliability of the deci-
49
Page 4


AnIntegratedCloud-basedFramework
forMobilePhoneSensing
Rasool Fakoor
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
rasool.fakoor@mavs.uta.edu
Mayank Raj
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
mayank.raj@mavs.uta.edu
Azade Nazi
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
azade.nazi@mavs.uta.edu
Mario Di Francesco
Dept. of Comp. Sci. and Eng.
Aalto University
Espoo, Finland
mario.di.francesco@aalto.?
Sajal K. Das
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
das@uta.edu
ABSTRACT
Nowadays mobile phones are not only communication de-
vices, but also a source of rich sensory data that can be
collected and exploited by distributed people-centric sensing
applications. Among them, environmental monitoring and
emergency response systems can particularly bene?t from
people-based sensing. Due to the limited resources of mo-
bile devices, sensed data are usually o?oaded to the cloud.
However, state-of-the art solutions lack a uni?ed approach
suitable to support diverse applications, while reducing the
energy consumption of the mobile device. In this paper, we
speci?cally address mobile devices as rich sources of multi-
modal data collected by users. In this context, we propose
an integrated framework for storing, processing and deliv-
ering sensed data to people-centric applications deployed in
the cloud. Our integrated platform is the foundation of a
new delivery model, namely, Mobile Application as a Ser-
vice (MAaaS), which allows the creation of people-centric
applications across di?erent domains, including participa-
tory sensing and mobile social networks. We speci?cally
address a case study represented by an emergency response
system for ?re detection and alerting. Through a prototype
testbed implementation, we show that the proposed frame-
work can reduce the energy consumption of mobile devices,
while satisfying the application requirements.
Categories and Subject Descriptors
C.2.4 [COMPUTER-COMMUNICATION
NETWORKS]: Distributed Systems—Distributed applica-
tions;C.2.1[COMPUTER-COMMUNICATION
NETWORKS]: Network Architecture and Design —Dis-
tributed networks
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for pro?t or commercial advantage and that copies
bear this notice and the full citation on the ?rst page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior speci?c
permission and/or a fee.
MCC’12, August 17, 2012, Helsinki, Finland.
Copyright 2012 ACM 978-1-4503-1519-7/12/08 ...$15.00.
General Terms
Design, Reliability
Keywords
Mobile Cloud Computing, People-centric Applications, Mo-
bile Application as a Service (MAaaS), Publish/Subscribe,
Mobile Phone Sensing
1. INTRODUCTION
The proliferation of mobile devices, combined with their
diverse sensing capabilities, has made them a ubiquitous
and rich source of sensory data. Such data include loca-
tion (e.g., GPS), orientation (e.g., gyroscope, accelerome-
ter), textual (social) and multimedia (e.g., image, audio and
video) data [9]. They can be jointly used to develop a wide
range of applications aimed at improving the quality of life.
These applications include environmental [12, 3] and tra?c
monitoring [18, 17], navigation [13] and emergency response
systems [10], as well as smart healthcare and well-being [9].
Many of the applications mentioned above are people-
centric, inthesensethattheyassistindividualusers, agroup
of people, or even a large community, by leveraging sen-
sory data from a large number of sources [2]. Depending on
user demands, sensing mechanisms can be either passive or
active, namely, participatory. Even though some of these
applications have been deployed in many environments [2,
9], integrating them into a single platform is still an open
research problem. This is due to the presence of a large
number of applications with contrasting requirements, di-
verse sensing sources and platforms, as well as multi-modal
data.
In this paper, we propose an integrated framework which
allows uni?ed sensing for diverse people-centric applications.
To ensure scalability, availability and robustness of the sens-
ing and application environments, the framework is de?ned
as part of the cloud infrastructure itself. Speci?cally, our in-
tegrated platform is the foundation of a new delivery model
– namely, Mobile Application as a Service (MAaaS) – which
allows the creation of people-centric applications across dif-
ferent domains, including participatory sensing and mobile
social networks. Each mobile device acts as an indepen-
dent multi-modal data source, and the proposed cloud-based
47 48
tionshosted on the cloud and select those they are interested
in directly from the MCA.
The rest of the system components is part of the cloud
infrastructure. The User Adaptation Module (UAM) in-
terfaces the MCA with the Mobile Cloud Engine (MCE).
Speci?cally, there is one UAM for each MCA. When mul-
tiple MCEs share a common user (namely, a MCA), the
related device sends the data once to the UAM which, in
turn, sends a copy to the relevant MCEs.
The Storage Module (SM) allows to save and retrieve sen-
sorydata. Itsupportsmulti-modaldata, rangingfromscalar
data (e.g., location, accelerometer and gyroscope readings)
to generic textual information (e.g., summary descriptions
uploaded by the users) and multimedia data (i.e., images,
audio and videos captured through the mobile device). As
a consequence, the SM is data-independent, but provides
means to interconnect to the cloud for mining and knowl-
edge extraction purposes. The major features of the SM are
scalability and responsiveness.
The Application Interface Module (AIM) interacts with
the framework to get the required data. It further allows
the application developer to access computing and storage
resources on-demand for ?exibility and scalability.
Finally, the Mobile Cloud Engine (MCE) processes and
controls the data ?ow in the framework, and is instantiated
for each application. Since the MCE is the core component
of our solution, it will be detailed in a separate subsection
below.
3.1 Mobile Cloud Engine
Our proposed framework contains two types of mobile
cloud engines. The ?rst type is called control MCE or
(MCE0). There is only one MCE0 in the whole framework,
independent of the actual people-centric applications. The
MCE0 is responsible to initiate other MCEs when a new ap-
plication request arrives in the framework. In addition, it
always collects the sensory data from mobile users in pre-
de?ned periods, regardless of the application requests. The
data collected by the MCE0 can be used for di?erent major
purposes, namely: mining; processing; and as a baseline to
start other MCEs. The second class of MCEs is instanti-
ated based on the actual application requests. These MCEs
are mainly employed to deal with speci?c people-centric ap-
plications. However, both classes of MCEs have the same
structure to provide coherency in the framework.
The MCE is composed of three sub-modules. The pub-
lish/subscribe (PS) module is primarily responsible for es-
tablishing the data ?ow between UAMs and the MCEs. The
PS module is composed of control and data message queues.
The control message queue handles all the control messages,
such as the data sending rate. The data message queues are
composed of multiple queues, each associated with a di?er-
ent priority (e.g., low, medium, and high). The data queue
delivers the captured data from the UAM to the associated
applications, and also forwards them to the storage module.
Furthermore, the PS module delivers the processed infor-
mation back to the mobile users. In this framework, the PS
modules are classi?ed in three categories, according to the
related priorities, namely, low, medium, and high. Di?erent
priorities are mainly used to regulate the sampling rate at
the mobile device, the data reporting rate to the cloud, and
the actual data processing within the cloud.
Thedecisionmodule (DM)handlesandregulatesthetrans-
actions among the di?erent parts of the MCE. It also classi-
?es the PS modules based on the requests received from ap-
plications. Speci?cally, once an application request arrives
in the MCE, the decision module identi?es which type of
PS module is needed to satisfy the request. Furthermore, it
recognizes the potential group of mobile users who can par-
ticipate to the application query. Finally, the DM module
initiates the context-awareness module (described next) to
provide people-centric and location-based services. The de-
cision module provides functions to process and mine data.
It can use unsupervised learning approaches to deal with
unstructured data in the system. One example is to ?nd
a coherent group of mobile phone users who share common
interests (e.g., through their frequently visited places) or to
predict user behavior.
Finally, the context-awareness (CA) module provides con-
text information to the di?erent people-centric applications.
In its simplest form, the context is represented by location.
For example, a pollution monitoring application may con-
sider the pollutant concentration at the locations in the
vicinity of the user. In the more general case, context may
include the activity of the user or its social interactions with
other users.
4. CASE STUDY
Texas is very likely to experience wild?res in Summer.
The state of Texas and the federal government spend sub-
stantial amount of money to handle wild?res, in order to en-
sure public safety and reduce damages. For instance, from
November 2010 to October 2011, about 30,457 ?res spread
over 3,993,716 acres and destroyed 3,017 homes [16]. The
cost of damages to houses was estimated to be more than
100 million dollars [4]. Dispatching ?re ?ghters to the exact
location of wild?res and preventing the ?re from spreading
play an important role in e?ciently handling the emergency.
However, placingcamerasandsensorsatalllocationstocon-
tinuously collect information is not feasible due to the high
deployment and maintenance costs.
In this section, we demonstrate how our proposed archi-
tecture can help in this scenario. Let us assume that the
mobile cloud application for emergency response has already
been installed on the user mobile devices and that a ?re hap-
pens in an certain area. In the proposed framework, user
location is periodically updated and stored in the the mo-
bile cloud framework either explicitly by the mobile device
user or through a request from the wild?re monitoring and
response agency. When a wild?re is reported, the agency
initiates a request to the framework with the information of
the a?ected area. After receiving the request, the framework
?nds the people in that area, based on their GPS locations.
Then, the framework sends an emergency noti?cation and
requests for updates on the situation.
At the same time, the mobile cloud framework updates
the agency with current available data in the system un-
til new data arrives. The application can provide di?erent
visualizations, ranging from a map showing the location of
people who are in the vicinity to a live video stream of a cer-
tain place. By using this method, the agency can obtain an
accurate characterization of the wild?re without deploying
any camera or sensor. The application can provide indirect
communication between the agency and mobile phone users
to explicitly obtain additional information which, in turn,
can help increase the accuracy and reliability of the deci-
49
sions made by the agency. The work ?ow of the proposed
framework is described next.
4.1 Work ?ow
In the following, we will refer to an emergency response
system to illustrate the work ?ow between the di?erent com-
ponents of the framework. The simplest scenario is repre-
sented by the case where there are no applications send-
ing requests to the cloud. Recall from Section 3.1 that the
MCE0 periodically collects data from the registered users
regardless of the active applications. As a consequence, sen-
sory data from mobile devices are delivered to the appro-
priate UAM and then to a low priority queue. The MCE0
asynchronously polls the queue for new data. Furthermore,
the MCE0 processes the fetched data and forwards them to
the storage module.
Let us consider now the case where two agencies corre-
sponding to two di?erent applications – i.e., a ?re agency
and an emergency healthcare agency – intend to exploit the
cloud framework. They initiate two di?erent requests to the
system. Note that a new MCE instance is created for each
application. For instance, the ?re agency needs visual data
of the ?re or location of users in the vicinity of the location
of ?re, say X. Then a new MCE1 instance ?nds all users in
the vicinity of location X and sends them a noti?cation to
collect data relevant to the situation. On the other hand,
the emergency healthcare agency needs information about
physicians and nurses in the vicinity of location Y. To this
end, an additional MCE2 instance ?nds all the users in the
vicinity of location Y then sends them a noti?cation alert.
The step-by-step procedure for handling an individual re-
quest is illustrated in Figure 2. First, the agency initiates a
request to the corresponding MCE. Second, the MCE sends
a request to the storage module to ?nd all registered users
in the vicinity of the speci?ed location, e.g., X or Y.After
receiving the list of the matching users U = {u1,u2,..,un}
from the storage module, the MCE moves them from the low
priority queue to a higher priority queue, in order to collect
data at a higher rate. The MCS also sets up an individual
PS module for sending the data collection noti?cations to
the speci?c users in U. On the other side, the UAM polls
the PS module for noti?cations and sends the data collected
by the users to the relevant applications. Hence, if a user
subscribed to di?erent applications, sampled data are just
sent once, and the UAM forwards them to the correspond-
ing MCEs. Finally, the MCE polls the queue, fetches the
data, and hands them over to the CA and AIM modules,
e.g., for visualization purposes. Moreover, it stores these
data in the storage module. This procedure continues until
the agency requests the termination of the service. Once the
service is terminated, users are switched back to the lower
priority queues in the MCE0, and the higher priority queue
are released.
5. IMPLEMENTATION
We implemented a prototype system that o?ers the basic
functionalities of the proposed framework. The components
residing at the cloud have been realized by using Amazon
Web Services (AWS) [1] and Google Fusion Tables [6]. The
Mobile Cloud Application (MCA) was built for Android de-
vices running the 2.1 Platform [5] and the AWS SDK for
Android [1]. Additional details are provided below.
5.1 Mobile Cloud Engine (MCE)
The Publish/Subscribe (PS) module was implemented by
using Amazon’s Simple Noti?cation System (SNS) and Sim-
ple Queue Service (SQS) [1]. SNS allows a cloud application
to easily send noti?cations to the subscribers. SNS exploits
the concept of topic to broadcast a noti?cation from pub-
lishers to subscribers. In this context, the publisher is the
application that sends messages, while the subscriber is the
application (e.g., the one running on the mobile phone) that
can receive the messages. A topic is an access point which
de?nes speci?c subject or event type for publishing mes-
sages and allows clients to subscribe for noti?cations [1].
In other words, the topic is the connection point between
the sender and receivers. The noti?cation can be in the
form of SMS, email, SQS, and so on. In our framework, the
MCEcreatesthetopicandthemobileclientapplicationsub-
scribes (unsubscribes) the mobile phone to (from) the topic.
The framework uses the SQS as publisher and subscriber
to a topic. SQS is an asynchronous mechanism by which
the data or messages can be transferred between distributed
components even without losing messages or requiring each
component to be always available [1]. For example, the mo-
bile cloud application uploads the GPS data to the SQS and
the MCE fetches the data from the SQS when needed at
any time. SQS simpli?es the interactions between the mo-
bile cloud applications and the MCE. Furthermore, it sup-
ports read (and write) operations from (to) a large number
of mobile cloud applications. Moreover, it provides security
by authentication method. Using SQS in our framework
enables mobile phones to send a large number of messages
in parallel. In addition, SQS and SNS together enable the
MCE and mobile client applications to send messages to a
huge number of users in multiple formats quickly and easily.
When the MCE needs to send a message to a group of mo-
bile phones, it publishes the message to a topic. Once the
messages have been published to a topic, all the subscribers
of the SQS get the messages. Since, the mobile client ap-
plication of each phone regularly polls the SQS, once the
message is available in SQS, it can get the message.
The context-awareness module is realized with reference
to location-based information and is implemented through
the Google Fusion Tables API [6]. Google Fusion Tables
provide mechanisms for publishing, retrieving, and render-
ing location-based data. It has been chosen for its features
of scalability, accuracy, and availability, as well as for the
built-in visualization based on Google Maps. Google Fusion
tables are controlled by a Google account such that each ac-
count has its own fusion tables. In the proposed framework,
the MCE was connected to a Google account in order to be
able to use these features. Whenever the MCE has to return
the result to the application community in the form of maps,
it uses Google Fusion Tables. Speci?cally, the MCE writes
the result into a fusion table, and based on those data, ren-
ders the map by using the Google Fusion API. Once the map
has been generated, the MCE returns the obtained map and
displays it through the widget. In addition, the data in the
fusion table is continuously updated by MCE upon arrival
of new data. As a result, the module is capable of rendering
dynamic information to the map. In the current implemen-
tation, the decision module is limited to initialize the MCE,
instantiate the PS modules, and handle the transactions be-
tween other modules.
50
Page 5


AnIntegratedCloud-basedFramework
forMobilePhoneSensing
Rasool Fakoor
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
rasool.fakoor@mavs.uta.edu
Mayank Raj
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
mayank.raj@mavs.uta.edu
Azade Nazi
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
azade.nazi@mavs.uta.edu
Mario Di Francesco
Dept. of Comp. Sci. and Eng.
Aalto University
Espoo, Finland
mario.di.francesco@aalto.?
Sajal K. Das
Dept. of Comp. Sci. and Eng.
University of Texas at Arlington
Arlington, Texas, USA
das@uta.edu
ABSTRACT
Nowadays mobile phones are not only communication de-
vices, but also a source of rich sensory data that can be
collected and exploited by distributed people-centric sensing
applications. Among them, environmental monitoring and
emergency response systems can particularly bene?t from
people-based sensing. Due to the limited resources of mo-
bile devices, sensed data are usually o?oaded to the cloud.
However, state-of-the art solutions lack a uni?ed approach
suitable to support diverse applications, while reducing the
energy consumption of the mobile device. In this paper, we
speci?cally address mobile devices as rich sources of multi-
modal data collected by users. In this context, we propose
an integrated framework for storing, processing and deliv-
ering sensed data to people-centric applications deployed in
the cloud. Our integrated platform is the foundation of a
new delivery model, namely, Mobile Application as a Ser-
vice (MAaaS), which allows the creation of people-centric
applications across di?erent domains, including participa-
tory sensing and mobile social networks. We speci?cally
address a case study represented by an emergency response
system for ?re detection and alerting. Through a prototype
testbed implementation, we show that the proposed frame-
work can reduce the energy consumption of mobile devices,
while satisfying the application requirements.
Categories and Subject Descriptors
C.2.4 [COMPUTER-COMMUNICATION
NETWORKS]: Distributed Systems—Distributed applica-
tions;C.2.1[COMPUTER-COMMUNICATION
NETWORKS]: Network Architecture and Design —Dis-
tributed networks
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for pro?t or commercial advantage and that copies
bear this notice and the full citation on the ?rst page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior speci?c
permission and/or a fee.
MCC’12, August 17, 2012, Helsinki, Finland.
Copyright 2012 ACM 978-1-4503-1519-7/12/08 ...$15.00.
General Terms
Design, Reliability
Keywords
Mobile Cloud Computing, People-centric Applications, Mo-
bile Application as a Service (MAaaS), Publish/Subscribe,
Mobile Phone Sensing
1. INTRODUCTION
The proliferation of mobile devices, combined with their
diverse sensing capabilities, has made them a ubiquitous
and rich source of sensory data. Such data include loca-
tion (e.g., GPS), orientation (e.g., gyroscope, accelerome-
ter), textual (social) and multimedia (e.g., image, audio and
video) data [9]. They can be jointly used to develop a wide
range of applications aimed at improving the quality of life.
These applications include environmental [12, 3] and tra?c
monitoring [18, 17], navigation [13] and emergency response
systems [10], as well as smart healthcare and well-being [9].
Many of the applications mentioned above are people-
centric, inthesensethattheyassistindividualusers, agroup
of people, or even a large community, by leveraging sen-
sory data from a large number of sources [2]. Depending on
user demands, sensing mechanisms can be either passive or
active, namely, participatory. Even though some of these
applications have been deployed in many environments [2,
9], integrating them into a single platform is still an open
research problem. This is due to the presence of a large
number of applications with contrasting requirements, di-
verse sensing sources and platforms, as well as multi-modal
data.
In this paper, we propose an integrated framework which
allows uni?ed sensing for diverse people-centric applications.
To ensure scalability, availability and robustness of the sens-
ing and application environments, the framework is de?ned
as part of the cloud infrastructure itself. Speci?cally, our in-
tegrated platform is the foundation of a new delivery model
– namely, Mobile Application as a Service (MAaaS) – which
allows the creation of people-centric applications across dif-
ferent domains, including participatory sensing and mobile
social networks. Each mobile device acts as an indepen-
dent multi-modal data source, and the proposed cloud-based
47 48
tionshosted on the cloud and select those they are interested
in directly from the MCA.
The rest of the system components is part of the cloud
infrastructure. The User Adaptation Module (UAM) in-
terfaces the MCA with the Mobile Cloud Engine (MCE).
Speci?cally, there is one UAM for each MCA. When mul-
tiple MCEs share a common user (namely, a MCA), the
related device sends the data once to the UAM which, in
turn, sends a copy to the relevant MCEs.
The Storage Module (SM) allows to save and retrieve sen-
sorydata. Itsupportsmulti-modaldata, rangingfromscalar
data (e.g., location, accelerometer and gyroscope readings)
to generic textual information (e.g., summary descriptions
uploaded by the users) and multimedia data (i.e., images,
audio and videos captured through the mobile device). As
a consequence, the SM is data-independent, but provides
means to interconnect to the cloud for mining and knowl-
edge extraction purposes. The major features of the SM are
scalability and responsiveness.
The Application Interface Module (AIM) interacts with
the framework to get the required data. It further allows
the application developer to access computing and storage
resources on-demand for ?exibility and scalability.
Finally, the Mobile Cloud Engine (MCE) processes and
controls the data ?ow in the framework, and is instantiated
for each application. Since the MCE is the core component
of our solution, it will be detailed in a separate subsection
below.
3.1 Mobile Cloud Engine
Our proposed framework contains two types of mobile
cloud engines. The ?rst type is called control MCE or
(MCE0). There is only one MCE0 in the whole framework,
independent of the actual people-centric applications. The
MCE0 is responsible to initiate other MCEs when a new ap-
plication request arrives in the framework. In addition, it
always collects the sensory data from mobile users in pre-
de?ned periods, regardless of the application requests. The
data collected by the MCE0 can be used for di?erent major
purposes, namely: mining; processing; and as a baseline to
start other MCEs. The second class of MCEs is instanti-
ated based on the actual application requests. These MCEs
are mainly employed to deal with speci?c people-centric ap-
plications. However, both classes of MCEs have the same
structure to provide coherency in the framework.
The MCE is composed of three sub-modules. The pub-
lish/subscribe (PS) module is primarily responsible for es-
tablishing the data ?ow between UAMs and the MCEs. The
PS module is composed of control and data message queues.
The control message queue handles all the control messages,
such as the data sending rate. The data message queues are
composed of multiple queues, each associated with a di?er-
ent priority (e.g., low, medium, and high). The data queue
delivers the captured data from the UAM to the associated
applications, and also forwards them to the storage module.
Furthermore, the PS module delivers the processed infor-
mation back to the mobile users. In this framework, the PS
modules are classi?ed in three categories, according to the
related priorities, namely, low, medium, and high. Di?erent
priorities are mainly used to regulate the sampling rate at
the mobile device, the data reporting rate to the cloud, and
the actual data processing within the cloud.
Thedecisionmodule (DM)handlesandregulatesthetrans-
actions among the di?erent parts of the MCE. It also classi-
?es the PS modules based on the requests received from ap-
plications. Speci?cally, once an application request arrives
in the MCE, the decision module identi?es which type of
PS module is needed to satisfy the request. Furthermore, it
recognizes the potential group of mobile users who can par-
ticipate to the application query. Finally, the DM module
initiates the context-awareness module (described next) to
provide people-centric and location-based services. The de-
cision module provides functions to process and mine data.
It can use unsupervised learning approaches to deal with
unstructured data in the system. One example is to ?nd
a coherent group of mobile phone users who share common
interests (e.g., through their frequently visited places) or to
predict user behavior.
Finally, the context-awareness (CA) module provides con-
text information to the di?erent people-centric applications.
In its simplest form, the context is represented by location.
For example, a pollution monitoring application may con-
sider the pollutant concentration at the locations in the
vicinity of the user. In the more general case, context may
include the activity of the user or its social interactions with
other users.
4. CASE STUDY
Texas is very likely to experience wild?res in Summer.
The state of Texas and the federal government spend sub-
stantial amount of money to handle wild?res, in order to en-
sure public safety and reduce damages. For instance, from
November 2010 to October 2011, about 30,457 ?res spread
over 3,993,716 acres and destroyed 3,017 homes [16]. The
cost of damages to houses was estimated to be more than
100 million dollars [4]. Dispatching ?re ?ghters to the exact
location of wild?res and preventing the ?re from spreading
play an important role in e?ciently handling the emergency.
However, placingcamerasandsensorsatalllocationstocon-
tinuously collect information is not feasible due to the high
deployment and maintenance costs.
In this section, we demonstrate how our proposed archi-
tecture can help in this scenario. Let us assume that the
mobile cloud application for emergency response has already
been installed on the user mobile devices and that a ?re hap-
pens in an certain area. In the proposed framework, user
location is periodically updated and stored in the the mo-
bile cloud framework either explicitly by the mobile device
user or through a request from the wild?re monitoring and
response agency. When a wild?re is reported, the agency
initiates a request to the framework with the information of
the a?ected area. After receiving the request, the framework
?nds the people in that area, based on their GPS locations.
Then, the framework sends an emergency noti?cation and
requests for updates on the situation.
At the same time, the mobile cloud framework updates
the agency with current available data in the system un-
til new data arrives. The application can provide di?erent
visualizations, ranging from a map showing the location of
people who are in the vicinity to a live video stream of a cer-
tain place. By using this method, the agency can obtain an
accurate characterization of the wild?re without deploying
any camera or sensor. The application can provide indirect
communication between the agency and mobile phone users
to explicitly obtain additional information which, in turn,
can help increase the accuracy and reliability of the deci-
49
sions made by the agency. The work ?ow of the proposed
framework is described next.
4.1 Work ?ow
In the following, we will refer to an emergency response
system to illustrate the work ?ow between the di?erent com-
ponents of the framework. The simplest scenario is repre-
sented by the case where there are no applications send-
ing requests to the cloud. Recall from Section 3.1 that the
MCE0 periodically collects data from the registered users
regardless of the active applications. As a consequence, sen-
sory data from mobile devices are delivered to the appro-
priate UAM and then to a low priority queue. The MCE0
asynchronously polls the queue for new data. Furthermore,
the MCE0 processes the fetched data and forwards them to
the storage module.
Let us consider now the case where two agencies corre-
sponding to two di?erent applications – i.e., a ?re agency
and an emergency healthcare agency – intend to exploit the
cloud framework. They initiate two di?erent requests to the
system. Note that a new MCE instance is created for each
application. For instance, the ?re agency needs visual data
of the ?re or location of users in the vicinity of the location
of ?re, say X. Then a new MCE1 instance ?nds all users in
the vicinity of location X and sends them a noti?cation to
collect data relevant to the situation. On the other hand,
the emergency healthcare agency needs information about
physicians and nurses in the vicinity of location Y. To this
end, an additional MCE2 instance ?nds all the users in the
vicinity of location Y then sends them a noti?cation alert.
The step-by-step procedure for handling an individual re-
quest is illustrated in Figure 2. First, the agency initiates a
request to the corresponding MCE. Second, the MCE sends
a request to the storage module to ?nd all registered users
in the vicinity of the speci?ed location, e.g., X or Y.After
receiving the list of the matching users U = {u1,u2,..,un}
from the storage module, the MCE moves them from the low
priority queue to a higher priority queue, in order to collect
data at a higher rate. The MCS also sets up an individual
PS module for sending the data collection noti?cations to
the speci?c users in U. On the other side, the UAM polls
the PS module for noti?cations and sends the data collected
by the users to the relevant applications. Hence, if a user
subscribed to di?erent applications, sampled data are just
sent once, and the UAM forwards them to the correspond-
ing MCEs. Finally, the MCE polls the queue, fetches the
data, and hands them over to the CA and AIM modules,
e.g., for visualization purposes. Moreover, it stores these
data in the storage module. This procedure continues until
the agency requests the termination of the service. Once the
service is terminated, users are switched back to the lower
priority queues in the MCE0, and the higher priority queue
are released.
5. IMPLEMENTATION
We implemented a prototype system that o?ers the basic
functionalities of the proposed framework. The components
residing at the cloud have been realized by using Amazon
Web Services (AWS) [1] and Google Fusion Tables [6]. The
Mobile Cloud Application (MCA) was built for Android de-
vices running the 2.1 Platform [5] and the AWS SDK for
Android [1]. Additional details are provided below.
5.1 Mobile Cloud Engine (MCE)
The Publish/Subscribe (PS) module was implemented by
using Amazon’s Simple Noti?cation System (SNS) and Sim-
ple Queue Service (SQS) [1]. SNS allows a cloud application
to easily send noti?cations to the subscribers. SNS exploits
the concept of topic to broadcast a noti?cation from pub-
lishers to subscribers. In this context, the publisher is the
application that sends messages, while the subscriber is the
application (e.g., the one running on the mobile phone) that
can receive the messages. A topic is an access point which
de?nes speci?c subject or event type for publishing mes-
sages and allows clients to subscribe for noti?cations [1].
In other words, the topic is the connection point between
the sender and receivers. The noti?cation can be in the
form of SMS, email, SQS, and so on. In our framework, the
MCEcreatesthetopicandthemobileclientapplicationsub-
scribes (unsubscribes) the mobile phone to (from) the topic.
The framework uses the SQS as publisher and subscriber
to a topic. SQS is an asynchronous mechanism by which
the data or messages can be transferred between distributed
components even without losing messages or requiring each
component to be always available [1]. For example, the mo-
bile cloud application uploads the GPS data to the SQS and
the MCE fetches the data from the SQS when needed at
any time. SQS simpli?es the interactions between the mo-
bile cloud applications and the MCE. Furthermore, it sup-
ports read (and write) operations from (to) a large number
of mobile cloud applications. Moreover, it provides security
by authentication method. Using SQS in our framework
enables mobile phones to send a large number of messages
in parallel. In addition, SQS and SNS together enable the
MCE and mobile client applications to send messages to a
huge number of users in multiple formats quickly and easily.
When the MCE needs to send a message to a group of mo-
bile phones, it publishes the message to a topic. Once the
messages have been published to a topic, all the subscribers
of the SQS get the messages. Since, the mobile client ap-
plication of each phone regularly polls the SQS, once the
message is available in SQS, it can get the message.
The context-awareness module is realized with reference
to location-based information and is implemented through
the Google Fusion Tables API [6]. Google Fusion Tables
provide mechanisms for publishing, retrieving, and render-
ing location-based data. It has been chosen for its features
of scalability, accuracy, and availability, as well as for the
built-in visualization based on Google Maps. Google Fusion
tables are controlled by a Google account such that each ac-
count has its own fusion tables. In the proposed framework,
the MCE was connected to a Google account in order to be
able to use these features. Whenever the MCE has to return
the result to the application community in the form of maps,
it uses Google Fusion Tables. Speci?cally, the MCE writes
the result into a fusion table, and based on those data, ren-
ders the map by using the Google Fusion API. Once the map
has been generated, the MCE returns the obtained map and
displays it through the widget. In addition, the data in the
fusion table is continuously updated by MCE upon arrival
of new data. As a result, the module is capable of rendering
dynamic information to the map. In the current implemen-
tation, the decision module is limited to initialize the MCE,
instantiate the PS modules, and handle the transactions be-
tween other modules.
50
PS
Module
Decision
Module
Find users for Q
UAM
Message
Queues
Storage
Module
U = { u 1 , u 2 ,   , u n }
Add users U
A request Q is 
received via 
the AIM
Setup
Poll
Send notification
Poll
Send data
Send data
Save Data, User IDs
Data are 
sampled at the 
mobile device
Repeated 
periodically{
Data are sent 
to the CA and 
AIM modules
Figure 2: Work ?ow within the system components for a given request
5.2 Storage Module
Inthecurrentimplementation,dataaredividedintoscalar
and multimedia data (as de?ned in Section 3) and managed
accordingly. In detail, the framework uses two di?erent stor-
age modules: Amazon SimpleDB and Amazon Simple Stor-
age Service (S3). SimpleDB is a non-relational database
which provides high availability and simplicity in adminis-
tration [1]. SimpleDB is used to store scalar (textual) data
such as sensory readings. When a scalar data arrives at the
MCE, they are associated with the corresponding user ID,
which is part of the message sent by the mobile cloud appli-
cation, and then stored in SimpleDB. The idea behind using
a non-relational database is to reduce the database admin-
istration overhead and provide fast access to the data [1].
When a mobile phone has multimedia data, the MCA stores
them into Amazon Simple Storage Service (S3), and sends
amessagecontainingareferencetothedatainS3toSQS.
S3 can be used to store and retrieve any amount of data [1].
By using S3 and SimpleDB together, we use the features of
SimpleDBasregulardatabaseaswellasthestoragecapacity
of S3 for multimedia data.
5.3 User Adaptation Module
Data arriving at the UAM are forwarded to the relevant
MCE(s). Speci?cally, when multiple MCEs have common
users (i.e., in terms of the corresponding MCAs), the data
captured by the MCAs are forwarded to all MCEs. To this
end, the UAM implementation is based on two components:
a ?lter process and a queue. The ?lter process forwards the
data to the concerned MCE, while the queue is used by the
?lter process to forward the data to multiple MCEs.
6. EXPERIMENTAL RESULTS
In order to characterize the advantages of using our frame-
work, we carried out two sets of experiments and studied
the impact of the uni?ed sensing approach in terms of en-
ergy consumption of the mobile device as well as scalability.
In the ?rst set of experiments, we assumed that the pro-
posed framework is deployed in the cloud and is exploited
by 4 applications. A thin mobile client installed on the mo-
bile device collectively sampled and uploaded data to the
UAM on behalf of the 4 applications. In the second set of
experiments, we did not exploit our framework, but rather
installed individual applications on the mobile device. We
varied the number of applications from 1 to 4. They sam-
pled and uploaded data individually on the mobile device.
In all the experiments, data was captured and uploaded to
the cloud every second.
The evaluation was conducted by using a Google Nexus
One smartphone. The duration of the experiment was set
to 1,000 seconds. We measured the energy consumed by
the mobile device for uploading messages in each experi-
ment through the PowerTutor energy pro?ler application
[19]. The obtained results are shown in Figure 3. Each
data point represents the average energy consumption over
10 iterations of each experiment, while the error bars rep-
resent the related con?dence interval with a 95% con?dence
level. Figure 3 shows that the energy consumption of the
mobile devices increase with the number of applications.
Even though di?erent applications require the same data,
such as GPS coordinates or images, the device separately
samples the data for each of them. Hence, as a result, it
leads to same data being separately sent to each application.
Thus, the mobile device experiences higher energy consump-
tion with the increase in the number of applications. In con-
trast, the same data required by multiple applications are
captured and sent to the UAM only once in our framework.
The MCE collects and processes the data from each UAM
and forwards it to the relevant application. Thus, this re-
sults in a lower energy consumption for the device, as it is
not interacting with each application individually.
51
Read More
Offer running on EduRev: Apply code STAYHOME200 to get INR 200 off on our premium plan EduRev Infinity!

Content Category

Related Searches

video lectures

,

pdf

,

shortcuts and tricks

,

Exam

,

AnIntegratedCloud-basedFrameworkforMobilePhoneSensingRasool FakoorDept. of Comp. Sci. and Eng. Computer Science Engineering (CSE) Notes | EduRev

,

Important questions

,

mock tests for examination

,

Summary

,

past year papers

,

Objective type Questions

,

Extra Questions

,

AnIntegratedCloud-basedFrameworkforMobilePhoneSensingRasool FakoorDept. of Comp. Sci. and Eng. Computer Science Engineering (CSE) Notes | EduRev

,

ppt

,

Previous Year Questions with Solutions

,

Free

,

MCQs

,

practice quizzes

,

Sample Paper

,

Semester Notes

,

AnIntegratedCloud-basedFrameworkforMobilePhoneSensingRasool FakoorDept. of Comp. Sci. and Eng. Computer Science Engineering (CSE) Notes | EduRev

,

study material

,

Viva Questions

;