See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/262056565

Developing Vehicular Data Cloud Services in the IoT Environment

Article  in  IEEE Transactions on Industrial Informatics · May 2014

DOI: 10.1109/TII.2014.2299233

CITATIONS READS
259 1,176

3 authors, including:

Wu He Guoqiang Yan
Old Dominion University 247 PUBLICATIONS   4,629 CITATIONS   
127 PUBLICATIONS   3,955 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Gencyber Summer Camp View project

All content following this page was uploaded by Wu He on 16 February 2017.

The user has requested enhancement of the downloaded file.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TII.2014.2299233, IEEE Transactions on Industrial Informatics
Developing Vehicular Data Cloud Services in the
IoT Environment
Wu He, Gongjun Yan and Li Da Xu, Senior Member, IEEE
Abstract— The advances in cloud computing and Internet of
Things (IoT) have provided a promising opportunity to resolve the
challenges caused by the increasing transportation issues. We
present a novel multi-layered vehicular data cloud platform by
using cloud computing and IoT technologies. Two innovative
vehicular data cloud services: an intelligent parking cloud service
and a vehicular data mining cloud service for vehicle warranty
analysis in the IoT environment are also presented. Two modified
data mining models for the vehicular data mining cloud service: a
Naïve Bayes model and a Logistic Regression model are presented
in details. Challenges and directions for future work are also
provided.
Index Terms—Cloud Computing, Service-Oriented
Architecture (SOA), Internet of Things (IoT), Intelligent
Transportation Systems, Automobile Service
I. INTRODUCTION
M odern vehicles are increasingly equipped with a large
amount of sensors, actuators and communication
devices (mobile devices, GPS devices, embedded computers).
In particular, numerous vehicles have possessed powerful
sensing, networking, communication and data processing
capabilities and can communicate with other vehicles or
exchange information with the external environments over
various protocols include HTTP, TCP/IP, SMTP, WAP and
NGTP (Next Generation Telematics Protocol) [1]. As a result,
many innovative telematics services [2] such as remote security
for disabling engine and remote diagnosis have been developed
to enhance drivers’ safety, convenience and enjoyment.
The advances in cloud computing and IoT have provided a
promising opportunity to further address the increasing
transportation issues such as heavy traffic, congestion and
vehicle safety. In the past few years, researchers have proposed
a few models that use cloud computing for implementing
intelligent transportation systems (ITS). For example, a new
vehicular cloud architecture called ITS-Cloud was proposed to
Manuscript received September 12, 2013. Accepted for publication January
2, 2014.
Copyright © 2009 IEEE. Personal use of this material is permitted.
However, permission to use this material for any other purposes must be
obtained from the IEEE by sending a request to pubs-permissions@ieee.org
Wu He is with Old Dominion University, Norfolk, VA 23529, USA
(phone:757-683-5008; email:whe@odu.edu)
Gongjun Yan is with the University of Southern Indiana, Evansville, IN
47712, USA (gyan@usi.edu)
Li Da Xu is with the Institute of Computing Technology, Chinese Academy
of Sciences; Shanghai Jiao Tong University; University of Science and
Technology of China, and Old Dominion University, Norfolk, VA 23529, USA
(email:lxu@odu.edu).
1551-3203 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1
improve vehicle-to-vehicle communication and road safety [3].
A cloud-based urban traffic control system was proposed to
optimize traffic control [4]. Based on a service-oriented
architecture, this system uses a number of software services
(SaaS) such as Intersection Control Services, Area Management
Service, cloud service discovery service and Sensor Service to
perform different tasks. These services also interact with each
other to exchange information and provide a solid basis for
building a collaborative traffic control and processing system in
a distributed cloud environment.
As an emerging technology caused by rapid advances in
modern wireless telecommunication, IoT has received a lot of
attention and is expected to bring benefits to numerous
application areas including health care, manufacturing and
transportation [5, 6, 7, 8]. Currently, the use of IoT in
transportation are still in its early stage and most research on
intelligent transportation systems has not leveraged the IoT
technology as a solution or an enabling infrastructure. To this
end, we propose to use both cloud computing and IoT as an
enabling infrastructure for developing a Vehicular Data Cloud
platform where transportation-related information such as
traffic control and management, car location tracking and
monitoring, road condition, car warranty and maintenance
information can be intelligently connected and made available
to drivers, auto-makers, part-manufacturer, vehicle quality
controller, safety authorities and regional transportation
division. An experiment of using data mining models to analyze
vehicular data clouds in the IoT environment was also
conducted to demonstrate the feasibility of vehicular data
mining service.
The rest of the paper is organized as follows. In section II, we
provide a brief review of vehicular networks, cloud computing
in the automotive domain, and IoT in the automotive domain. In
section III, we propose a novel multi-layered vehicular data
cloud platform using existing cloud computing and IoT
technologies. Section IV presents two innovative vehicular data
cloud services: an intelligent parking cloud service and a
vehicular data mining cloud service for vehicle warranty
analysis in the IoT environment. Two modified data mining
models for the vehicular data mining cloud service: a Naïve
Bayes model and a Logistic Regression model are presented in
details. Challenges and directions for future work are given in
Section V. Section VI presents our conclusions.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TII.2014.2299233, IEEE Transactions on Industrial Informatics
II. RELATED WORK
A. Vehicular networks
Wireless technology leads to the development of vehicular
networks in the past decades. The original idea is that the
roadside infrastructure and the radio-equipped vehicles could
communicate using wireless networks. To make networking
operations such as routing more effective, researchers had
developed a dynamic inter-vehicle network called Vehicular
Ad-hoc Networks (VANET). VANETs were primarily designed
to support the communication between different vehicles (V2V)
and the communication between vehicles and the roadside
infrastructures (V2I) [9]. VANETs possess hybrid architecture
and integrate ad hoc networks, wireless LAN and cellular
technology [10] for intelligent transportation system (ITS).
Furthermore, many VANET applications were developed by
numerous vehicle manufacturers, government agencies and
industrial organizations. Initially most VANET applications
were focused on improving drivers’ safety and offered functions
such as traffic monitoring and update, emergency warning and
road assistance [11]. In recent years, many non-safety-related
VANET applications such as entertainment and gaming
applications have been developed.
B. Cloud Computing in the Automotive Domain
Cloud Computing has been proposed to reshape vehicular
software and services in the automotive domain. As more and
more cars are equipped with devices that can access the Internet,
Olariu, Khalil, and Abuelela [11] propose to integrate existing
vehicular networks, various sensors, on-board devices in
vehicles, and cloud computing to create vehicular clouds. They
suggest that vehicular clouds are technologically feasible and
will have a significant impact on the society once they are built.
Thus, both existing automobile software and a variety of
information resources are being virtualized and packaged as
services to build vehicular clouds. Different vehicular services
are often combined and used to implement the mapping,
encapsulation, aggregation and composition and allow vehicles
to interact with various hosted services outside the vehicles.
Currently, using the modular approach, multi-layer and
service-oriented architectures to integrate various vehicular
resources and services appears to be the most promising model
and framework for building vehicular cloud service platforms.
By using the modular approach to decompose a complex system
into smaller subsystems according to their functions, we can
divide a vehicular cloud service platform into a number of
functional services and subsystems such as traffic
administration, service routing, information processing, vehicle
warranty analysis and mining, and so on. As cloud computing
includes three distinct services - platform as a service (PaaS),
infrastructure as a service (IaaS) as well as the popular software
as a service (SaaS), a compound of SaaS, PaaS and IaaS should
be leveraged for building vehicular cloud service platforms.
Furthermore, clouds can also be divided into private, public and
hybrid clouds. Thus, vehicular cloud service platforms can also
be designed to be a hybrid cloud where some services such as
1551-3203 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
2
user information query services can be hosted on public cloud
platforms and other missing-critical services such as traffic
administration should be hosted on private cloud platforms [12].
A taxonomy was developed to classify VANET-related clouds
into the following three types: Vehicles using Clouds, Vehicular
Clouds, and Hybrid Clouds [13].
Multi-layer approaches and service-oriented architecture [14,
15, 16] have been proposed as the main architecture to construct
various vehicular cloud service platforms. Iwai and Aoyama [1]
propose to develop a cloud service system for automobiles
(a.k.a., the DARWIN system) using SOA (Service-Oriented
Architecture) as an enabling architecture [17, 18]. DARWIN
contains key service components such as Service Process
Manager and Service Space and these components interact with
various services both inside and outside of vehicles to form a
comprehensive vehicular cloud. DARWIN also provides
protocols to support interoperability between existing vehicular
software and cloud-based services. Wang, Cho, Lee and Ma
[19] propose a vehicle cloud computing architecture composed
of three functional tiers: cloud service, communication and
device tiers. By using cloud computing techniques such as SOA,
the three-layer architecture allows heterogeneous devices,
network and services to exchange information and collaborate
in a real time manner. A three-layer V-Cloud architecture was
proposed [10] to combine vehicular cyber-physical systems
with cloud computing technologies to offer essential services
for drivers. The V-Cloud architecture includes three layers:
in-car vehicular cyber-physical system, V2V network, and V2I
network. Each layer has numerous sub-components. The
ITS-Cloud proposed by Bitam & Mellouk [3] includes three
layers: cloud layer, communication layer and end-users layer. In
particularly, the cloud layer was divided into both static and
dynamic cloud to support different services needed by various
stakeholders of the vehicular clouds. A new architecture named
VehiCloud was developed to transform traditional vehicular
networks into a service-oriented cloud architecture [20]. By
taking advantage of emerging cloud computing technologies
[21, 22, 23], VehiCloud has been implemented and tested to
address V2V communication issues and extend the capabilities
of embedded devices and mobile devices though road
experiments.
C. Internet of Things in the Automotive Domain
The integration of sensors and communication technologies
provides a way for us to track the changing status of an object
through the Internet. IoT explains a future in which a variety of
physical objects and devices around us such as various sensors,
radio frequency identification (RFID) tags, GPS devices, and
mobile devices will be associated to the Internet and allow these
objects and devices to connect, cooperate and communicate
within social, environmental, and user contexts to reach
common goals [24,25]. As an emerging technology, the Internet
of Things (IoT) is expected to offer promising solutions to
transform transportation systems and automobile services in the
automobile industry. Speed & Shingleton [26] propose an idea
to use the “unique identifying properties of car registration
plates” to connect various things. As vehicles have increasingly
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TII.2014.2299233, IEEE Transactions on Industrial Informatics
3

powerful sensing, networking, communication and data and primarily hosts highly dynamic vehicular applications
processing capabilities, IoT technologies can be used to harness which may have issues running on the conventional clouds [26].
these capabilities and share under-utilized resources among For example, traffic-related applications and smart parking
vehicles in the parking space or on the road. For example, IoT applications are suitable for the temporary cloud. The
technologies make it possible to track each vehicle’ existing temporary cloud often needs to communicate with the
location, monitor its movement and predict it future location. conventional clouds and there is a frequent exchange of data and
By integrating with cloud computing, wireless sensor services between the two clouds [13]. Based on the layered
network, RFID Sensor Networks, satellite network and other architecture in Figure 1, heterogeneous IoT-related devices,
intelligent transportation technologies, a new generation of network, community technologies and cloud-based
IoT-based vehicular data clouds can be developed and deployed services on different layers can be integrated to exchange
to bring many business benefits such as predicting increasing information, share resources and collaborate on the
road safety, reducing road congestion, managing traffic, and
clouds.
recommending car maintenance or repair. Some preliminary
work of using IoT technologies to improve intelligent
transportation systems has been conducted in recent years. For
example, an intelligent informatics system (iDrive system)
developed by BMW used various sensors and tags to monitor
the environment such as tracking the vehicle location and the
road condition to provide driving directions [27]. Leng and
Zhao [12] propose an intelligent internet-of-vehicles system
(known as IIOVMS) to collect traffic information from the
external environments on an ongoing basis and to monitor and
manage road traffic in real time. Lumpkins [28] discusses how
intelligent transportation systems could use IoT devices in the
vehicle to connect to the cloud and how numerous sensors on
the road could be virtualized to leverage the processing
capabilities of the cloud. Qin, Long, Zhang, and Huang [27]
propose a technology architecture that uses cloud computing,
IoT and middleware technologies to enable the innovation of Figure 1. An architecture for IoT-based vehicular data clouds
automobile services. Zhang, Chen, and Lu [29] designed an
intelligent monitoring system to track the location of The proposed IoT-based vehicular data cloud platform supports
refrigerator trucks using IoT technologies. three new cloud services as indicated in Table 1 [30]:
New Services Description
III. PROPOSED A VEHICULAR DATA CLOUD PLATFORM IN THE Network and Data Vehicles provide their networking and data
IOT ENVIRONMENT Processing as a processing capabilities to other vehicles
Service, i.e. through the cloud.
Figure 1 shows the layered architecture of our proposed Infrastructure As A
IoT-based vehicular data cloud platform. By integrating various Service (IAAS)
devices such as sensors, actuators, controllers, GPS devices, Storage as a Service Some vehicles may need specific
mobile phones and other Internet access equipments and (SAAS) applications that require large amount of
employing networking technologies (wireless sensor network, storage space. Thus, vehicles that have
cellular network, satellite network and others), cloud unused storage space can share their storage
computing, IOT, and middleware, this platform supports V2V space as a cloud-based service.
Platform as a Service As a community, vehicular data clouds offer
and V2I communication mechanism and is able to collect and
(PAAS) a variety of cooperative information services
exchange data among the drivers, vehicles, and roadside such as traffic information, hazardous
infrastructure such as cameras and street lights. The goal of this location warning, lane change warning and
platform is to provide real-time, economic, secure and parking availability.
on-demand services to customers through the associated clouds
Table 1. Novel services for IoT-based vehicular data clouds
including a conventional cloud and a temporary cloud
(vehicular cloud) [3]. The conventional cloud is composed of
In this proposed layered architecture, different layers have
virtualized computers and provides SaaS, PaaS, and IaaS to
different purposes. In general, the layers on the bottom provide
interested customers. For example, cloud management services
a foundational support for the layers on the top. SOA will be
and many traffic administration applications can be hosted on
applied to integrate different information and communication
the conventional cloud. The temporary cloud is typically formed
services and connect in-vehicle and out-vehicle applications
on demand and is composed of under-utilized computing,
seamlessly though the vehicular data clouds. SOA allows
networking and storage facilities of vehicles and is designed to
vehicular application developers to organize, aggregate and
expand the conventional cloud in order to increase the whole
package applications into new business applications services.
cloud’s computing, processing and storing capabilities. The
As a mature technology for enterprise application integration,
temporary cloud supports a compound of SaaS, PaaS, and IaaS

1551-3203 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TII.2014.2299233, IEEE Transactions on Industrial Informatics
4

service-oriented architecture (SOA) provides guidelines to parked with cars, the light will show green color; otherwise, the
integrate heterogeneous Web services, applications and light shows red color. These sensors connect to computer center
different middleware systems. Middleware is used to hide the to report the status of every parking slot on an ongoing basis.
implementation details of underlining technologies and
provides support for the integration of specific applications
deployed on the vehicular data cloud [24]. By leveraging the
SOA-based and IoT-based vehicular data cloud platform,
innovative services can be developed by car manufacturers,
government agencies and third-party service provides. In the
section, we propose two innovative vehicular data cloud
services.
Figure 3. Vacancy Detections by Sensors
IV. VEHICULAR DATA CLOUD SERVICES We also designed an infrastructure to publish advertisement
from the parking lot (see Figure 4). There are wireless
In this section, we are interested in introducing two vehicular transceiver tower in the parking lot and multiple transceivers
data cloud services as examples of PAAS shown in Section III. (shown as si to si+3) installed on the roadside. The wireless
One service is intelligent parking cloud service and the other tower in the parking lot can obtain vacant slot information from
one is the mining vehicular maintenance data service the computer center where the status of the parking lot is
A. Intelligent Parking Cloud Service constantly monitored. Therefore, wireless tower can broadcast
the parking lot information and parking plan as business
Finding available parking space is challenging in many cities
strategies for economical benefits. As the wireless transmission
and often leads to issues such as congestion, road accidents
range is limited, we have roadside transceivers to relay the
and psychological frustration. To make it easier to find available
parking information to remote areas, as shown in Figure 4.
parking space, an intelligent parking cloud service that collects
and analyzes geographic location information, parking
availability information, parking space reservation and order
information, traffic information and vehicle information though
sensor detection and the clouds is needed. Using a modular
approach, a software architecture [31] for implementing the
intelligent parking cloud service is proposed below (Figure 2).

Figure 2. Software architecture for intelligent parking cloud

service
Each vehicle is pre-enlisted with a transceiver with short
transmission range (about 1m) and a processor with simple
computing capacity. The transceiver can be common devices,
such as zigbee, bluetooth devices and infrared devices with low
cost. Both the processor and the wireless transceiver are enlisted
into an EDR. We designed a parking lot with WIFI network,
infrared devices and parking belts to detect misparked cars.
When a car enters the parking lot and heads to the reserved
parking slot, the entrance booth will validate the reservation. If
the parking spot is validated, a direction-related guidance will
be uploaded to the car for finding the reserved spot. The infrared
device, lights and parking belt will work together to detect and
prevent misparking. As shown in Figure 3, the Bluetooth
communication will be activated when the front wheel presses
the belt-a. The TRD and belt-a in Figure 3 will validate
reservation confirmation as necessary. We use an infrared
device to validate whether the car is parked instead of using the
slot for a temporary purpose. If the parking slot is correctly
Figure 4. Parking Cloud Service
B. Intelligent Parking Service Models
In this section, the parking process has been modeled as a
birth-death stochastic process. The parking revenue could be
predicted by using such a model. The birth and the death of
parking mean that a vehicle enters and exits a parking slot
respectively at time t. We were able to obtain the birth and death
rate by using traffic detectors or other sensors [31]. We assume
that there is a huge number of parking slots and this number can
be considered infinite for practical purposes. Let X(t) to be the
number of slots in occupied at time t. We write {X(t);t≥0} is a
birth and death process. Therefore, we can read for all t>0,h>0
and i=0,1,.... The occupied slot at time t is

1551-3203 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TII.2014.2299233, IEEE Transactions on Industrial Informatics
5

where λ is the birth rate and µ is the death rate. We note the
probability of a car parking event that occurs in (t,t+h) is
independent from the number of occupied parking slots at time
t. If p (t) refers to the probability when the number of parking
j
cars at time t is j, we can derive the probability of p (t) n
j The joint probability p(c,w ,⋯,w )∝p(c)
1 n  p(wi|c)
in
i=1
which we assume p(w |c,w )=p(w |c) , i.e. w is independent of
i j i i
each other word w when j≠i. In reality, this assumption is not
j
always true but this assumption can greatly simplify the problem
Based on p (t), we can compute the mean value and reach acceptable accuracy in real problems.
j
Therefore, we write
n
p(c)  p(w |c)
i
i=1
p(c|w ,⋯,w )= .
C. Mining Vehicular Maintenance Data Service 1 n  n 
Another application we are interested in the proposed  p(c')  p(wi|c')
vehicular data cloud is to mine maintenance data. c'  i=1 
Maintenance of vehicles are frustrated and heavily loaded for We made efforts to find the “best" class of Naïve Bayes
drivers. More importantly, auto-manufacturers and auto-parts classification. The best class is typically the maximum of a
designer and manufacturers also desperately seek feedback posteriori (MAP) class
from end users to improve the quality of produces and c =argmax p(c|x)
max c∈C
enhance competition capability to the foreign auto-makers. n
We can merge maintenance data from both users and =argmax p(c)  p(w |c)
repair-people. c∈C i
i=1
The merged data are nature language text descriptions of But we usually compute in practice by taking the log:
maintenance. We place these texts into main files. To dig out  n 
auto-parts warranty information, we adopt nature language c
max
=argmax
c∈C 
logp(c)+  logp(w |c)
i
text mining technologies to these merged texts. As an  i=1 
emerging research area, currently limited studies have been Each conditional parameter logp(w |c) is used as a weight to
published regarding how data mining techniques could be i
applied to vehicular networks or clouds. Few models were indicate the usefulness of an indicator w to the class c. The
i
developed and tested for mining vehicular data collected from prior log p(c) is also used as a weight to indicate the relative
vehicular networks or data clouds. In this sub-section, we frequency of c. The sum of the term weights and the log prior is
present a modified Naïve Bayes model [32] and a Logistic used to measure the amount of available evidence for the
Regression model which have been adopted in our research of document that is located in the class c. We are going to pick the
implementing vehicular data mining cloud service which is class that has the largest summation.
introduced in next section. We obtain p(x|c) and p(c) from the training data set:
Given a document space X={x ,⋯,x } where all N
1 d c
p(c)=
documents are represented in this space. We use x for X=x . A N
i where N is number of docs in class c and N is total number of
set of fixed classes is marked as C={c ,c ,⋯,c } . We use c c
1 2 m documents. The conditional probabilities can be calculated:
for C=c and a training set D={(x ,c ),...,(x ,c )} of T
i 1 1 n n wi
classified documents. Each labeled document <x ,c >∈X×C. A p(w |c)=
i i
classifier is to map documents to classes: γ:X→C . In this
i  Tt'
section, we present an optimized model called Naïve Bayes t'∈D
Classifier. T refers to the number of words w in the training documents
wi i

1) Naïve Bayes Classifier D from the class c.  T means the total number of tokens in
t'
In the Naïve Bayes model, we calculated the probability of a t'∈D
document x locating in the class c by using Bayes rules: training document.
p(c)p(x|c)
p(c|x)= 2) Optimization (1)
p(x)
A document x is a list of words: x={w ,⋯,w } . Thus, In this section, we are interested in the optimization problem of
1 n Naïve Bayes Classifier. Let each document be represented by a

1551-3203 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TII.2014.2299233, IEEE Transactions on Industrial Informatics
6

T words of x. The Logistic regression, a discriminative model,

word count vector x=(t ,⋯,t ) . We assume for each class c,
1 v estimates p(c|x) directly.
the probability distribution of a document follows the We also put an emphasis on the learning classifiers, c=f(x),
multinomial distribution with parameter Θ : for a set of training data sets {(x ,c ),...,(x ,c )} . For a text
c 1 1 n n
v tw T
document, the vector x =(t ,⋯,t ) consist of transformed
p(x|c)∝  Θc (7)i 1 d
w=1 w word frequencies t ,i∈d from the training document. The values
i
The log likelihood is c ∈[−1,+1] serve as class labels that encode the
T i
logp(x|c)=x logΘ +const. (8)
membership (+1) or non-membership (-1) of the vector in the
c
We also assume that the multinomial distribution assume category of c . To map x to real number, we calculate the inner
i
conditional independence of feature dimensions 1,⋯,v given the v
class c. Given a training set {(x ,c ),...,(x ,c )} . Our task, product between x and the parameter vector Θ∈R :
1 1 n n T
in this step, is to find the best parameters Θ x
τ={p(c=j),Θ ,...,Θ } . Therefore, we translate the Specifically, we calculated the conditional probability in
1 v logistic link function form:
n xw 1
model as p(x|c=j)∝  Θj . According to the p(c=+1|x,Θ) =
T
w=1 w 1+exp(−Θ x)
maximum-likelihood estimation (MLE), the maximum of the Let there are m classes {c ,c ,⋯,c } . For a class, the
1 2 m
joint (log) likelihood of the training set: probability of the class can be defined:
* T
cmax=logp((x,y) |τ);skip τ below exp(Θk x)
1:n
n p(c=k|x)=
m T
=log  p(x ,y )  exp(Θi x)
i i
i=1 i=1
n Therefore, we are looking for Θ that minimizes the
=  logp(y )logp(x |y ) following expression:
i i i
i=1 n
2 T
* min λ||Θ|| +  log(1+exp(−y Θ x ))
Therefore cmax is a constrained optimization problem. We Θ i i
i=1
m The optimization value of the log likelihood loss form can
write  p(c=j)=1,m is the number of classes and typically be solved by Newton-Raphson iterations (or by
j=1 iterative reweighted least squares for logistic regression) [34,
v 35].
 Θj =1,∀j=1⋯m It is easy to solve it using Lagrange Recall that both Naïve Bayes and Logistic Regression are
w=1 w linear classifiers. They both divide the documents X with a
multipliers [33] and arrive at hyperplane. But they differ from each other: Naïve Bayes
m optimizes a generative objective function, while Logistic
 [ci=j] Regression optimizes a discriminative objective function. In
i=1 practice, logistic regression often has higher accuracy when
p(c=j)= training data set is large and on the other hand Naïve Bayes has
m
and an advantage when the size of the training data set is small.
 xiw D. Vehicular Data Mining Cloud Service
i:c =j As vehicular data clouds contain a variety of heterogeneous
i
Θ = data and information resources, effective data mining service
jw v
  xiu must be developed to quickly detect dangerous road situations,
i:c =ju=1 issue early warning messages, and assist drivers to make
i informed decisions to prevent accidents [36]. Data mining
The above results are intuitively explained as following: they services can also be used to assess drivers’ behavior or
are class frequency in the training data set, and the word performance of vehicles to find problems in advance. The core
frequency of each class. of any data mining service is the data mining models [37, 38].
So far few models were developed and tested for mining
3) Logistic Regression Model vehicular data collected from vehicular networks or data clouds.
Logistic regression model is also used for both classifying and Below is a specialized data mining service for car warranty
clustering [34]. As a generative model, Naïve Bayes can model early-warning analysis. We applied the models we developed in
the joint p(x,c), with the assumption of independence on the the above section to design and develop the data mining service.

1551-3203 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TII.2014.2299233, IEEE Transactions on Industrial Informatics
In vehicle manufacturing process, sometimes some quality
issues can be hidden for a long time without being identified.
Due to a lack of events or signals to correlate several discrete
issues, potential problems may not be investigated at all. To
avoid accidents, it is important to develop new techniques that
reveal these hidden problems in advance. By using the two
modified data mining models (Naive Bayes Classifier and
Logistic Regression Classifier) to cluster and classify the real
car warranty and maintenance data we collected from a local
automobile company, we demonstrated how data mining cloud
service could be used to identify potential issues that could
become a problem later. This experiment assumes a new
product is under development and has some potential but
unknown issues. As a result of applying the two data mining
models, we were able to acquire some preliminary results (see
Figure 5). We found that the precision p in column cross
dropped dramatically. The drop is associated with the k’s value
which is 3. In other words, the model found that there are three
groups among all the corpus of natural language text. We show
the group distribution in Figure 6.
Figure 5. Clustering results for unknown warranty issues
Figure 6. Clusters of unknown warranty issues
V. CHALLENGES AND DIRECTIONS FOR FUTURE WORK
IoT-based vehicular data clouds must be efficient, scalable,
secure and reliable before they could be deployed at a
large-scale. Existing algorithms and mechanisms are
unsatisfactory to meet all these requirements at the same time.
Below is a description of some of these challenges.
 Scalability and Technology Integration. The effectiveness
of a vehicular cloud depends on its scalability to handle a
dynamically changing number of vehicles. In addition to
handling regular traffic, vehicular clouds must be able to
handle traffic spike or sudden demands caused by special
events or situations such as sport games or emergencies.
More development on optimization algorithms that
coordinate virtual machines, storage space and network
bandwidth to balance server workload and improve
1551-3203 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
7
computing resource utilization on the vehicular clouds is
needed [39]. As new devices and technologies are coming
out each year, developing effective IoT Middleware that
supports integration of these new technologies and devices
[40] with existing in-vehicle technologies from automobile
manufacturers will be a challenge.
 Performance, reliability and quality of service. As vehicles
are often on the move, the vehicular networking and
communication is often intermittent or unreliable. More new
mechanisms are needed to enhance the communication
reliability with reduced traffic overhead. For example, Chen
et al. [41] developed a new transmission protocol to make
the conventional Zigbee protocol more reliable. Cross layer
data synchronization mechanisms should also be designed to
minimize the traffic overhead between layers. Acceleration
data compression algorithms for resource-constrained
sensors, actuators and other Internet-access devices need
improvement to be able to effectively and efficiently
compress a large amount of raw data generated. Multiple
processing units and support in different cloud data centers
are needed to minimize the service response time, improve
availability and stability of the service, and increase cloud
reliability and fault tolerance. Real-time evaluation
mechanisms regarding the performance, reliability and
service quality on vehicular data clouds will have to be
further developed.
 Security and privacy. There are some security and privacy
concerns with vehicular data clouds due to a lack of
established infrastructure for authentication and
authorization [39]. A low security level of vehicular data
clouds is unacceptable for vehicular services regarding
transportation safety. For example, roadside attackers may
maliciously send many fake requests to the parking cloud
service and reserve many parking spaces. They can also send
misleading parking availability information or wrong
location information to the parking cloud service to cause
chaos. Trust relationships are hard to be built in vehicular
clouds because of the large and dynamically changing
number of vehicles on the road. Balanced security measures
are needed to enhance the security and trust of cloud
services without limiting the flexibility of the system. In
addition, many drivers don’t want their vehicle locations to
be tracked or monitored due to the worries about their
privacy. Reasonable efforts in technology [42], law and
regulation are needed to secure the vehicular data clouds and
prevent unauthorized access to or disclosure of the privacy
data. For example, implementing security authentication in
vehicular data clouds is required security countermeasure.
 Lack of global standards for device and service integration,
security, privacy, architecture, and communications. Global
standards are essential to avoid conflicts between locally
developed vehicular data clouds [27]. However, as there are
a number of stakeholders involved in vehicular data clouds
and complex dependencies among these stakeholders also
exist, it is challenging to establish global standards to lower
the complexity and make vehicular data clouds more
compatible and cost-effective. Further efforts on
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TII.2014.2299233, IEEE Transactions on Industrial Informatics
standardization are needed to coordinate various efforts and
resources for implementing vehicular data clouds.
VI. CONCLUSION
In this article, we present a novel modular and multi-layered
vehicular data cloud platform based on cloud computing and
IoT technologies. We also discuss how cloud services could be
developed to make the vehicular data clouds useful. This study
makes contributions by proposing a novel software architecture
for the vehicular data clouds in the IoT environment, which has
the capabilities to integrate numerous devices available within
vehicles and devices in the road infrastructure.
IoT-based Vehicular Data Clouds are expected to be the
backbone of future intelligent transportation systems with the
ultimate goal of making driving safer and more enjoyable.
However, research on integrating IoT with the vehicular data
clouds is still in its infancy and existing study on this topic is
highly insufficient. To make vehicular data clouds useful,
numerous services such as road navigation, traffic management,
remote monitoring, urban surveillance, information and
entertainment, and business intelligence [43-47] need to be
developed and deployed on vehicular data clouds. A number of
challenges such as security, privacy, scalability, reliability,
quality of service, and lack of global standards still exist. Due to
the complexity involved in implementing vehicular clouds and
integrating various devices and systems with vehicular clouds
[48-51], a systematic approach and collaboration among
academia, the automobile companies, law enforcement,
government authorities, standardization groups and cloud
service providers are need to address these challenges. Though
with many challenges, IoT and cloud computing provide
tremendous opportunities for technology innovation in the
automobile industry [52, 53] and will serve as enabling
infrastructures for developing vehicular data clouds.
References
[1] A. Iwai, and M. Aoyama, "Automotive cloud service systems based on
service-oriented architecture and its evaluation", Proceedings of IEEE
International Conference on Cloud Computing, pp. 638-645, 2011.
[2] Y. Zhao, “Telematics: safe and fun driving”, IEEE Intelligent systems,
17(1), pp. 10-14, 2002
[3] S.Bitam, and A. Mellouk,"ITS-cloud: Cloud computing for Intelligent
transportation system," In IEEE Global Communications Conference
(GLOBECOM), pp. 2054-2059, 2012.
[4] P.Jaworski, T.Edwards, J. Moore, and K. Burnham, "Cloud computing
concept for Intelligent Transportation Systems,” Proceedings of the 14th
International IEEE Conference on Intelligent Transportation Systems (ITSC),
pp. 391-936, 2011.
[5] Z. Pang, L. Zheng, J. Tian, S. Kao-Walter, E. Dubrova, and Q. Chen,
“Design of a terminal solution for integration of in-home health care devices
and services towards the Internet-of-Things”, Enterprise Information Systems,
(ahead-of-print), 1-31, 2013.
[6] L. Xu, “Enterprise Systems: State-of-the-Art and Future Trends,” IEEE
Transactions on Industrial Informatics, VOL. 7, NO. 4,pp. 630-640, 2011.
[7] S. Li, L. Xu, and X. Wang, “Compressed sensing signal and data acquisition
in wireless sensor networks and internet of things,” IEEE Transactions on
Industrial Informatics, 9(4), 2177-2186, 2013.
[8] Y. Li, M. Hou, H. Liu, and Y. Liu, “Towards a theoretical framework of
strategic decision, supporting capability and information sharing under the
context of Internet of Things,” Information Technology and Management,
13(4), 205-216, 2012.
1551-3203 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
8
[9] P. Papadimitratos, A. La Fortelle, K. Evenssen, R. Brignolo, and S.
Cosenza, “Vehicular communication systems: Enabling technologies,
applications, and future outlook on intelligent transportation,” IEEE
Communications Magazine, 47(11), 84-95, 2009.
[10] H. Abid, L.T.T. Phuong, J. Wang, S. Lee, and S. Qaisar, “V-Cloud:
vehicular cyber-physical systems and cloud computing”, Proceedings of the
4th International Symposium on Applied Sciences in Biomedical and
Communication Technologies (p. 165), 2011.
[11] S. Olariu, I. Khalil, and M. Abuelela, “Taking VANET to the clouds,”
International Journal of Pervasive Computing and Communications, 7(1),
pp.7-21, 2011
[12] Y. Leng, and L. Zhao, "Novel design of intelligent internet-of-vehicles
management system based on cloud-computing and Internet-of-Things,"
Proceedings of 2011 International Conference on Electronic and Mechanical
Engineering and Information Technology (EMEIT), Vol. 6, pp. 3190-3193,
2011.
[13] R. Hussain, J. Son, H. Eun, S. Kim, and H. Oh, "Rethinking Vehicular
Communications: Merging VANET with cloud computing," Proceedings of
IEEE 4th International Conference on Cloud Computing Technology and
Science (CloudCom), pp. 606-609, 2012.
[14] R. Mietzner, F. Leymann, and T. Unger, “Horizontal and vertical
combination of multi-tenancy patterns in service-oriented application,”
Enterprise Information Systems, vol. 5, no. 1, pp. 59-77, 2011.
[15] H. Panetto, and J. Cecil, “Information systems for enterprise integration,
interoperability and networking: theory and applications,” Enterprise
Information Systems, 7(1), 1-6, 2013.
[16] S. Hachani, L. Gzara, and H. Verjus, “A service-oriented approach for
flexible process support within enterprises: application on PLM systems,”
Enterprise Information Systems, 7(1), 79-99, 2013.
[17] F. Tao, H. Guo, L. Zhang, and Y. Cheng, “Modelling of combinable
relationship-based composition service network and the theoretical proof of its
scale-free characteristics,” Enterprise Information Systems, 6(4), 373-404,
2012.
[18] D. Paulraj, S. Swamynathan, and M. Madhaiyan, “Process model-based
atomic service discovery and composition of composite semantic web services
using web ontology language for services (OWL-S),” Enterprise Information
Systems, 6(4), 445-471, 2012.
[19] J. Wang, J.Cho, S. Lee, and T. Ma, "Real time services for future cloud
computing enabled vehicle networks," Proceedings of 2011 International
Conference on Wireless Communications and Signal Processing (WCSP), pp.
1-5, 2011.
[20] Y. Qin, D. Huang, and X. Zhang (2012), “VehiCloud: Cloud Computing
Facilitating Routing In Vehicular Networks,” Proceedings of the IEEE 11th
International Conference on Trust, Security and Privacy in Computing and
Communications (TrustCom), pp. 1438-1445.
[21]H. Wang, W. He, and F. K. Wang, “Enterprise cloud service architectures,”
Information Technology and Management, 13(4), 445-454, 2012.
[22]Q. Li, Z. Y. Wang, W. H. Li, J. Li, C. Wang, and R. Y. Du, “Applications
integration in a hybrid cloud computing environment: modeling and platform,”
Enterprise Information Systems, 7(3), 237-271, 2012.
[23]L. Ren, L. Zhang, F. Tao, X. Zhang, Y.L. Luo, and Y. Zhang, “A
methodology towards virtualization-based high performance simulation
platform supporting multidisciplinary design of complex products,” Enterprise
Information Systems, 6(3), 2012, 267-290, 2012.
[24] European Commission Information Society. Internet of Things in 2020: a
Roadmap for the Future. Available from: www.iot-visitthefuture.eu
[25] L. Atzori, A. Iera, and G. Morabito, “The internet of things: A survey,”
Computer Networks, 54(15), pp. 2787-2805, 2010.
[26]C. Speed, and D. Shingleton, “An internet of cars: connecting the flow of
things to people, artefacts, environments and businesses,” In Proceedings of
the 6th ACM workshop on Next generation mobile computing for dynamic
personalised travel planning, pp. 11-12, ACM, 2012.
[27] E. Qin, Y. Long, C. Zhang, and L. Huang, “Cloud Computing and the
Internet of Things: Technology Innovation in Automobile Service,” In Human
Interface and the Management of Information. Information and Interaction for
Health, Safety, Mobility and Complex Environments (pp. 173-180). Springer
Berlin Heidelberg, 2013.
[28] W. Lumpkins, “The Internet of Things Meets Cloud Computing,” IEEE
Consumer Electronics Magazine, 2(2), pp. 47-51, 2013
[29] Y. Zhang, B. Chen, and X. Lu, “Intelligent Monitoring System on
Refrigerator Trucks Based on the Internet of Things,” In Wireless
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TII.2014.2299233, IEEE Transactions on Industrial Informatics
9

Communications and Applications, pp. 201-206. Springer Berlin Heidelberg,

2012. Wu He received the B.S. degree in Computer Science from
[30] S. Arif, S. Olariu, J. Wang, G. Yan, W. Yang, and I. Khalil, “Datacenter at DongHua University, China, in 1998, and the Ph.D. degree
the Airport: Reasoning about Time-Dependent Parking Lot Occupancy,” IEEE in Information Science from the University of Missouri,
Transactions on Parallel and Distributed Systems, 23(11), 2067-2080, 2012. USA, in 2006. His research interests include Enterprise
[31] G. Yan, W. Yang, D. B. Rawat, and S. Olariu, “SmartParking: A secure Applications, Data Mining, Cyber Security and Knowledge
and intelligent parking system,” IEEE Intelligent Transportation Systems Management.
Magazine, 3(1), 18-30, 2011.
[32] D. Lowd, and P. Domingos, "Naive Bayes models for probability
estimation," In Proceedings of the 22nd international conference on Machine
learning (pp. 529-536). ACM, 2005. Gongjun Yan received his Ph.D. in Computer Science from
[33] R.T. Rockafellar, “Lagrange multipliers and optimality,” SIAM review, Old Dominion University in 2010. Currently he is an
183-238, 1993. Assistant Professor of Computer Science at the University
[34] G. Qian, Y. Wu, and Q. Shao, “A procedure for estimating the number of of Southern Indiana. He applied mathematics models and
clusters in logistic regression clustering,” J. Classification, 26(2):183–199, computer simulations to analyze research problems in
2009. information security, Vehicular Ad Hoc Networks, and
[35] T. J. Hastie, and D. Pregibon, Generalized linear models. Wadsworth & Wireless Communications.
Brooks/Cole, 1992.
[36] M. Faezipour, M. Nourani, A. Saeed, and S. Addepalli, “Progress and
challenges in intelligent vehicle area networks,” Communications of the ACM,
55(2), 90-100, 2012.
[37] L. Bulysheva, and A. Bulyshev, “Segmentation modeling algorithm: a Li Da Xu (M’86-SM’11) received the M.S. degree in
novel algorithm in data mining,” Information Technology and Management, information science and engineering from the University of
13(4), 263-271, 2012. Science and Technology of China, Hefei, China, in 1981,
[38] L. Zeng, L. Li, L. Duan, K. Lu, Z. Shi, M. Wang, W. Wu, and P. Luo, “ and the Ph.D. degree in systems science and engineering
Distributed data mining: a survey,” Information Technology and Management, from Portland State University, Portland, OR, USA, in
13(4), 403-409, 2012. 1986. Dr. Xu serves as the Founding Chair of IFIP TC8
[39] G. Yan, D. Wen, S. Olariu, and M. C. Weigle, “Security challenges in WG8.9 and the Founding Chair of the IEEE SMC Society
vehicular cloud computing,” IEEE Transactions on Intelligent Transportation Technical Committee on Enterprise Information Systems.
Systems, 14(1), 284-294, 2013.
[40]S. Li, L. Xu, X. Wang, and J. Wang, “Integration of hybrid wireless
networks in cloud services oriented enterprise information systems,” Enterprise ACKNOWLEDGMENT
Information Systems, 6(2), 165-187, 2012.
[41]S.K. Chen, T. Kao, C.T. Chan, C.N. Huang, C. Y. Chiang, C.Y. Lai, T.H. This work was supported in part by the National Natural Science Foundation of
Tung, and P.C. Wang, “A reliable transmission protocol for ZigBee-based China (NNSFC) under Grant 71132008, and U.S. National Science Foundation
wireless patient monitoring,” IEEE Transactions on Information Technology under Grant SES-1318470 and 1044845
in Biomedicine, 16(1):6-16, 2012.
[42] L. Li, “Technology designed to combat fakes in the global supply chain,”
Business Horizons, 56(2), 167-177, 2013.
[43] L. Duan, and L.D. Xu, “Business Intelligence for Enterprise Systems: A
Survey,” IEEE Transactions on Industrial Informatics, vol.8, no.3,
pp.679-687, 2012.
[44] M. Gerla, "Vehicular Cloud Computing," In The 11th IEEE Annual
Mediterranean Ad Hoc Networking Workshop (Med-Hoc-Net), pp. 152-155,
2012.
[45] S. Fang, L. Xu, H. Pei, Y. Liu, Z. Liu, Y. Zhu, J. Yan, and H. Zhang, “An
integrated approach to snowmelt flood forecasting in water resource
management,” IEEE Transactions on Industrial Informatics, 2013,
10.1109/TII.2013.2257807
[46]L. Zeng, L. Li, and L. Duan, “Business intelligence in enterprise computing
environment,” Information Technology and Management, 13(4), 297-310,
2012.
[47]J. E. Ingvaldsen, and J. A. Gulla, “Industrial application of semantic
process mining,” Enterprise Information Systems, 6(2), 139-163, 2012.
[48]Y. Xing, L. Li, and Z. Bi, ”Operations research (OR) in service industries:
a comprehensive review,” Systems Research and Behavioral Science, 30,
300-353, 2013.
[49]J. Wan, and J. D. Jones, “Managing IT service management
implementation complexity: from the perspective of the Warfield version of
systems science,” Enterprise Information Systems, 7(4), 490-522, 2011.
[50]S. Wang, L. Li, K. Wang, and J. D. Jones, “e-Business systems integration:
a systems perspective,” Information Technology and Management, 13(4),
233-249, 2012.
[51]L. D. Xu, “Introduction: Systems Science in Industrial Sectors,” Systems
Research and Behavioral Science, 30(3), 211-213, 2013.
[52] G. Goggin, “Driving the Internet: Mobile Internets, Cars, and the Social,”
Future Internet, 4(1), 306-321, 2012.
[53] J. Gubbi, R. Buyya, S. Marusic, and M. Palaniswami, "Internet of Things
(IoT): A vision, architectural elements, and future directions," Future
Generation Computer Systems, 29(7), pp. 1645-1660, 2013.

1551-3203 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
View publication stats http://www.ieee.org/publications_standards/publications/rights/index.html for more information.