IEEE Wireless Communications - April 2017 - page 76

IEEE Wireless Communications • April 2017
74
1536-1284/17/$25.00 © 2017 IEEE
Deve Manikantan Shila is
with United Technologies
Research Center.
Wenlong Shen and Yu Cheng
are with Illinois Institute of
Technology.
Xiaohua Tian is with Shanghai
Jiaotong University.
Xuemin (Sherman) Shen is
with University of Waterloo.
A
bstract
Cloud computing is a revolutionary paradigm
to deliver computing resources, ranging from data
storage/processing to software, as a service over
the network, with the benefits of efficient resource
utilization and improved manageability. The cur-
rent popular cloud computing models encompass
a cluster of expensive and dedicated machines to
provide cloud computing services, incurring sig-
nificant investment in capital outlay and ongoing
costs. A more cost effective solution would be to
exploit the capabilities of an
ad hoc cloud
which
consists of a cloud of distributed and dynamically
untapped local resources. The ad hoc cloud can
be further classified into static and mobile clouds:
an ad hoc static cloud harnesses the underuti-
lized computing resources of general purpose
machines, whereas an ad hoc mobile cloud har-
nesses the idle computing resources of mobile
devices. However, the dynamic and distributed
characteristics of ad hoc cloud introduce chal-
lenges in system management. In this article, we
propose a generic em autonomic mobile cloud
(AMCloud) management framework for automat-
ic and efficient service/resource management of
ad hoc cloud in both static and mobile modes.
We then discuss in detail the possible security and
privacy issues in ad hoc cloud computing. A gen-
eral security architecture is developed to facilitate
the study of prevention and defense approaches
toward a secure autonomic cloud system. This
article is expected to be useful for exploring
future research activities to achieve an autonomic
and secure ad hoc cloud computing system.
I
ntroduction
Cloud computing is a novel technology to deliver
computing resources, ranging from data storage
and processing to software, as a service over the
network, typically using Internet technologies. The
US National Institute of Standards and Technolo-
gy (NIST) has categorized cloud computing into
three service models [1]: software as a service,
platform as a service, and infrastructure as a ser-
vice. Still evolving, this revolutionary paradigm has
the potential provide several benefits, including
significant cost savings through increased operat-
ing and economic efficiencies, improved manage-
ability, and reduced maintenance. Furthermore, it
could also significantly boost partnership, agility,
and scalability, thus facilitating a truly global com-
puting model.
Nevertheless, existing archetypical cloud
computing models (public, private, or partner)
encompass a cluster of expensive and dedicated
machines to run those cloud computing resourc-
es, leading to significant investment in capital out-
lay and ongoing costs. It is further observed that
the resources inside the data centers often oper-
ate at “low utilization” due to resource stranding
and fragmentation. Moreover, the energy con-
sumed by machines housed in data centers also
embodies a financial burden on the organizations
that operate them as well as an infrastructure bur-
den on power utilities.
For a cost effective cloud computing mode, ad
hoc cloud computing aims to leverage untapped
local computing and storage resources to form
an ad hoc cloud of local resources. Based on the
nature of the local resources, we categorize ad
hoc cloud into two modes, ad hoc static cloud
and ad hoc mobile cloud. Ad hoc static cloud har-
nesses the underutilized computing resources of
general purpose machines owned by an enter-
prise, organization, or normal users. Though ad
hoc static cloud shares some of the characteristics
of the grid and volunteer computing, it incorpo-
rates new features, i.e. supporting diverse appli-
cations, rapid elasticity, and coordinated use of
computing resources for large number of users.
Ad hoc mobile cloud harnesses the idle comput-
ing resources of mobile devices owned by the
same or different individuals. The concept of ad
hoc mobile cloud is different from classic mobile
cloud models that allow offloading of mobile
applications to remote resource-rich clouds. In ad
hoc mobile cloud, a local pool of smart devices
(e.g. smartphones, Internet of things, etc.) will be
collected and aggregated to provide sufficient
resources for computationally expensive cloud
services. Note that ad hoc mobile cloud is closely
related to the concept of opportunistic comput-
ing, where a device can opportunistically lever-
age other available resources within the network.
Nevertheless, ad hoc mobile cloud targets a more
organized, large-scale resource sharing system
through resource virtualization and system man-
agement.
Ad hoc cloud is expected to provision a
D
evu
M
anikantan
S
hila
, W
enlong
S
hen
, Y
u
C
heng
,
X
iaohua
T
ian
,
and
X
uemin
(S
herman
) S
hen
AMC
loud
: T
oward
a
S
ecure
A
utonomic
M
obile
A
d
H
oc
C
loud
C
omputing
S
ystem
A
ccepted
from
O
pen
C
all
Digital Object Identifier:
10.1109/MWC.2016.1500119RP
1...,66,67,68,69,70,71,72,73,74,75 77,78,79,80,81,82,83,84,85,86,...132
Powered by FlippingBook