Trust Management for Vehicular Cloud Computing
Marcela Roxana Farcasescu
West University of Timisoara, Department of Informatics and Mathematics, Bd. Vasile Parvan, Timisoara, Romania
1 RESEARCH PROBLEM
In the last years, the business model of cloud
computing has spread viral in several domains,
promising to change the way we think about
computing and data storage. One of the biggest
concerns in cloud adoption since 2009, is considered
to be the trust offered from both perspective – cloud-
end user and cloud provider. A new and challenging
field that highlights the need of defining the trust
management concept is vehicular cloud computing.
Stephen Olariu and his team have define the
vision of vehicular cloud computing as a pool of
resources which need to be carefully handled by a
trusted interested party.
This is why the current paper is going to
combine two of the main challenges in Vehicular
Cloud Computing by creating trust models: the
authentication of the owner of the car and the
geolocation data sharing. The infotaiment unit of a
vehicle will soon have installed an accesible
operating system which is already used for our
mobiles, tablets, gadgets, etc. In this way, all the
current applications that can be installed on our
gadgets will soon be adapted to the infotaiment unit,
which means that the vehicular cloud will become a
necesity. With this in mind, the authentication of the
owner of a car will become an important security
issue, while the geolocation data sharing will remain
a critical privacy concern.
The innovative approach is to build trust models
based on driver’s behavior as the owner of his
vehicle, using the geolocation as a strategical item in
validating the user’s actions and requests. The
attacks simulation will consider the scenario of
loosing ownership of the vehicle and providing fake
geolocation data.
2 OUTLINE OF OBJECTIVES
The main activities of the research will be to:
Outline the security challenges in cloud
computing but also in vehicular cloud
environments
Mitigate the importance of a trust
management in a cloud environment
Build trust models for driver identification
and vehicle’s localization
Apply the trust models and analyse the results
Perform attack simulations: attacker as a
driver and fake geolocation received
Interpretation of the results by providing a
final conclusion on the confirmed or infirmed
theory
The research will be conducted from two main
directions: how to trust the information provided by
the driver and how to trust the services used in
communication (the information requested by the
service will be used only for obtaining the needed
data and not for other purposes).
Before getting into the exact details, the main
concepts used, need to be defined.
3 STATE OF THE ART
Concepts review:
Cloud Computing
Cloud computing combines in a successful recipe
the advantages of the grid computing with the
service-oriented. This introduces an elastic and rapid
provisioning service consumption model. Cloud
computing can overcome the running out of
capacity, capacity excess costs, tied-up capital. The
main characteristics of the cloud computing concept
are: on-demand self-service or pay-as-you-go, rapid
elasticity, resource pooling, measured service[Bernd
Grobauer]. In order to adopt cloud computing, a
cloud model should be chosen that will cover the
business needs (SaaS, PaaS, IaaS), the provider roles
and capabilities should be analysed, determine the
dependencies on the cloud provider, establish and
negotiate the SLA (Service Level Agreement), have
an exit strategy – in case the cloud provider will
disappear from the cloud ecosystem.
Cloud Security is a sub-domain of the computer
security that includes all the policies, controls,
14
Farcasescu, M.
Trust Management for Vehicular Cloud Computing.
In Doctoral Consortium (DCCLOSER 2016), pages 14-17
deployment techniques used in the infrastructure of
the cloud computing. Cloud based security refers to
the services present in the cloud environment. The
focus of this research paper relies on the sub-domain
of cloud based security.
Vehicular Cloud Computing
Vehicular cloud computing represent an extension of
the cloud computing paradigm describing: “a group
of largely autonomous vehicles whose corporate
computing, sensing, communication and physical
resources can be coordinated and dynamically
allocated to authorized users”[ Md Whaiduzzaman].
A shorter definition redesigns the concept of a
vehicle in the cloud computing context, describing it
as driver’s dependent entities, service providers and
communication tools. The differences between cloud
computing and vehicular cloud computing are
The main challenges of both of the paradigms are
establishing a reliable trust context for each vehicle
involved in the communication, but also to trust the
information provided by those entities. This is why
the trust management concept is required when
discussing about the security and privacy concerns
of the vehicular cloud computing.
Trust Management
Similar to the cloud computing, the trust concept has
multiple definitions, because it can be seen from
different perspectives: customer’s and cloud
vendor’s point of views. Trust is “the subjective
probability by which an individual, expects that
another individual, performs a given action on which
its welfare depends”[Talal H]. Trust represents
“more than the authorized nature of security
relations between human societies, from a stable and
healthy operation to a large extent thanks to the trust
relationship between the individuals, groups and
organizations.”[ R. K. L. Ko, P. Jagadpramana]
Marsh was one of the first researchers of the
computational trust. In the trust model that he
proposed as a direct interaction between agents, he
divided trust in three categories [Marsh S.]:
Basic Trust - based on experiences from the
past
General Trust - the trust that an agent has in
other agent
Situational Trust - the amount of trust that an
agent has in other agent in a specific situation
The bidirectional relationship between two
agents can be described at any moment by using the
degree of trust represented by the three categories.
Trust management represents the trust analysis
between agents using trust metrics that rely on
reputation, deception, persuasion and the optimal
decisions taken according to the trust level found.
The trust management techniques can be classified
in four main categories: policy, recommendation,
reputation, prediction. For each category, specific
trust models can be created. A trust model can be
described by a set of rules that reflect the
relationship between the cloud provider and
customer.
4 METHODOLOGY
4.1 Building the Trust Models
The trust models will be created as sets of rules
established by the cloud provider or end-cloud user
using a Rule Based Engine (for example the
Business Rule Engine from WSO2). The set of rules
will reflect a specific policy or recommendation.
The trust models will be used later on at the
authentication and geographical location
identification phases. The geo-location can be also
used as a second authentication factor.
From the authentication perspective, we would
like to try to focus on personal gadget (owned
device) usage as an authentication method in order
to gain access to the infortaiment (the entertainment
unit from the vehicle) data. If the owner of the car is
not recognized, the current driver should have
limited access to the information presented on the
infotaiment unit. The recognition will be done using
the sensors of the owned device which will
interpretate the user behaviour in the vehicle. On the
other side, geo-location can be used to confirm that
the driver behaviour.
Geo-location is a concept used in “information
systems security cycles to extrapolate the
geographical location of a subject (person or
system)”. There are several Location-Based Service
(LBS) applications on different platforms that
currently challenge the user’s trust levels and
privacy. The LBS application use the geolocation
data to obtain other information for the user. Usually
are installed on personal devices and help in
researching for specific points of interest (eg.
restaurants, museums, etc). The main difference
between Geolocation Services and Location-Based
Services is that the first ones don’t require an
accurate user address and the primary objective is to
protect the user’s privacy.
For the Geolocation model, the GPS data will be
collected using driver’s gadgets and the infotaiment
unit. After establishing a set of rules for each cloud
provider, a trust score will be associated considering
Trust Management for Vehicular Cloud Computing
15
the number of touched rules per policy (a rule will
be considered “touched” if the condition or criteria
used in the rule was fully respected).
4.2 Collecting the Data
The data will be collected using mobile phones,
tablets, other gadgets sensors. Usually the devices
have built-in sensors that measure motion,
orientation, and various environmental conditions.
The sensors will provide raw data with high
precision and accuracy, monitoring three-
dimensional device movement or positioning, or
ambient environment changes. The sensors can be
classified as: environmental, motion and position
sensors.
The purpose of the collected data is to be used as
an input for the trust models created in order to
approve or decline further access through the cloud
ecosystem.
After collecting the data, an heuristic approach
will be used in identifying similarities and use the
input data as a further authentication schemas.
4.3 Attack Simulations
The next steps will be to prove by simulating attacks
that the heuristics chosen are actually providing the
expected results and try to improve those to address
possible attack scenarios.
A. Stolen Device
A possible scenario can be considered when the
drivers gadget or owned device it’s now in attacker’s
hands. An attack graph will be created looking at the
chances of the attacker to gain full access to the
infotaiment data using the stolen device.
B. Fake Geolocation Data
Another possible scenario is when the attacker
which already stolen the device from the driver is
sending fake geolocation data. An attach graph will
be created to analyse all the possible scenarios.
4.4 Preparing for Vehicular Cloud
Computing Outages
The outage concept refers to the unavailability of a
service that usually it is a consequence of a server
loss of electricity power or malware
blockages[Nicholas Car]. Outages could be a serious
security vulnerability for a cloud ecosystem. When
an outage occurs in cloud computing, there are two
key factors: the type of outage and the time needed
for service recovery.
Outages can be scheduled or unexpected. If the
cloud provider is scheduling an outage for
administration reasons each month at the same date
time, the service outage can be easily tracked and
used in malicious scopes. An unexpected outage can
occur due to a malware attack that takes down the
DNS servers. When an outage is analyzed, security
issues must be considered in order to detect the root
causes. After an outage in cloud computing, the
cloud vendor should present the problem and explain
the identified issue. If this is not possible, the
viability of the cloud services will be usually
questioned. The focus of the cloud provider is to
prepare for outages, not to avoid them. Prediction
might also be beneficial, but it implies processing a
lot of monitoring data and might result in too many
false-positive alarms.
In particular for vehicular cloud computing, an
outage is actual represented by a missed
communication between the vehicle and the cloud
which means that the data from the vehicle to the
cloud is not transferred anymore or the vehicle
doesn’t receive anymore data from the cloud. This
situation can occur while driving in the areas where
the communication is not possible (depending on the
channel used).
5 EXPECTED OUTCOME
The expected outcome is to confirm the advantages
and disadvantages of using the trust management in
vehicular cloud environments by analysing the
obstacles and proposed -solutions. The expected
results will highlight the need of implementing the
trust models for the vehicular cloud computing
environment.
6 STAGE OF THE RESEARCH
A few attempts for building the trust models were
already perform in order to extract the difference
between what the customer provided and what the
services collect with or without the end-user’s
approval.
Currently, I am in the process of collecting the
data from several owned devices like tablets, mobile
phones, fitness braces. After collecting the data the
next step will be to implement the trust models
DCCLOSER 2016 - Doctoral Consortium on Cloud Computing and Services Science
16
which will contribute to the driver identification.
ACKNOWLEDGEMENTS
This work was supported by a grant of the Romanian
National Authority for Scientific Research and
Innovation, CNCS-UEFISCDI, project number PN-
II-RU-TE-2014-4-1501(2015-2017).
REFERENCES
Stephan Olariu, Gongjun Yan, Ding Wen, Michele
C.Weigle, 2012, “Security Challenges in Vehicular
Cloud Computing” IEEE Transactions on intelligent
transportation systems.
Md Whaiduzzaman, Mehdi Sookhak, Adbullah Gani,
Rajkumar Buyya, 2014, “A survey on vehicular cloud
computing”,Journal Network and Computer
Applications.
Mario Gerla, 2012, “Vehicular Cloud
Computing”,Vehicular Communications and
Applications Workshop.
Talal H. Noor, Quan Z.Sheng, Sherali Zeadally, Jian Yu,
2013, “Trust Management of Services in Cloud
Environments: Obstacles and Solutions”, ACM
Computing Surveys, Vol 46, No1, Article 12,
Publication Date: October 2013.
Bernd Grobauer, Tobias Walloschek, Elmar Stocker,
2010,”Understanding Cloud-Computing
Vulnerabilities”, Proceedings of the 2010 IEEE
Security and Privacy,pp. 1-14.
Cloud Security Alliance, August 23, 2012, “Cloud
Computing Vulnerability Incidents: a statistical
overview”, Cloud Vulnerabilities Working Group.
Josang, A., Keser, C., and Dimitrakos, T. , 2005.
"Can We Manage Trust?" In: P. Herrmann et al. (Eds.):
iTrust, LNCS 3477, 2005, pp. 93-107.
S. Marsh, 1994 “Formalising trust as computational
concept”, PhD Thesis, Department of Computer
Science, University of Stirling, 1994, pp.56-59.
R. K. L. Ko, P. Jagadpramana, M. Mowbray, S. Pearson,
M. Kirchberg, Q. Liang, et al.,2011 "TrustCloud - A
Framework for Accountability and Trust in Cloud
Computing," in IEEE 2nd Cloud Forum for
Practitioners (IEEE ICFP 2011), Washington DC,
USA, pp. 1-5.
Nicholas Carr, 2008,“The Big Switch - Rewiring the
world, from Edison to Google”, Second Edition, W.W.
Norton & Company Inc, New York, Part 1, Chapter 1,
pp. 14-25.
Trust Management for Vehicular Cloud Computing
17