XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE
Hybrid Quantum-Behaved Particle Swarm
Optimization for Mobile-Edge Computation
Offloading in Internet of Things
Shijie Dai
School of information science
and engineering
Xiamen University
Xiamen, China, 361005
xmu_dsj@sina.com
Minghui Liwang
School of information science
and engineering
Xiamen University
Xiamen, China,361005
minghuilw@stu.xmu.edu.cn
Yang Liu
School of information science
and engineering
Xiamen University
Xiamen, China, 361005
liuyangxmuce@126.com
Zhibin Gao*
School of information science
and engineering
Xiamen University
Xiamen, China,361005
gaozhibin@xmu.edu.cn
Lianfen Huang
School of information science
and engineering
Xiamen University
Xiamen, China, 361005
lfhuang@xmu.edu.cn
Xiaojiang Du
Department of Computer and
Information Sciences
Temple University
Philadelphi, PA, 19122, USA
dux@temple.edu
*Zhibin Gao is the corresponding author.
Abstract—Mobile edge computing (MEC) is a technology that
transfers resource to the edge of network, which spares more
attention to giving users easier access to network and computation
resources. Due to the large amount of data computation needed
for devices in Internet of Things, MEC technology is applied to
improve computing efficiency. Though MEC can be applied to the
Internet of Things, it needs further consideration on how to
efficiently and reasonably allocate computing resources, and how
to minimize the computing time of all users. This paper proposes
a computing resources allocation scheme based on hybrid
quantum-behaved particle swarm optimization. Simulation
experiments with the network environment based on the Internet
of Things is carried out. The results show that this algorithm can
accelerate the whole computing process and reduce the number of
iterations.
Keywords—Internet of things, Mobile edge computing,
offloading
I. INTRODUCTION
With the rapid development of information technology and
information industry, the Internet of Things (IoT), supported by
big data mining, cloud computing and machine learning, has
been applied to various fields. And the appearances of newly-
developing technology, cloud computing, bid data, AR and
other technologies, promote the industry IoT upgrading. A
recent study by NCTA in United States assumes that about 5.01
million Internet of Things will be connected to the Internet by
2020 [1].
The future era will rely on Internet technology to achieve the
intelligent life, covering the fields in home security,
environmental testing, energy, car networking, industrial
intelligence manufacturing and other. IoT transforms itself from
simple mode of things to things to the intelligent mode. IoT
typically involve a large number of smart sensors that sense
information from the environment and share it with the cloud
service for processing. IoT application services can be divided
into two types: one is a post hoc analysis, which collects data
through the IoT terminal. And it upload to the cloud through the
IoT private network or public network. Then the information
will combine with bid data to be filtered and analyzed in the
cloud. This application is often one-way. In other words, to
capture and analysis do not need feedback data transmission.
The other one is a real-time feedback type, which is making data
acquisition and analysis not only through the IoT terminal, but
also through the reverse real-time feedback. Such applications
have higher requirements for latency and reliability.
Currently the IoT architecture is still cloud-centric
architecture. Its main feature is that the communication exists
between terminal and the cloud, and the main type of application
services is a post hoc analysis. With the development of IoT,
real-time feedback application requirements will increase more
and more. And the IoT current architecture, cloud-centric
architecture, is clearly unable to meet the needs of such
applications. The main problems caused by generate data from
IoT are: 1) IoT application processing time may be limited by
the network delay in offloading data to the cloud. 2) the
generation and upload of a large amount of IoT data causes
network congestion resulting in further network delay.
To address the network problems designed in the Internet of
Things and similar applications, researchers have proposed to
bring computing closer to data generators and consumers. One
suggestion is the fog computation [2], which enables the device
to run the cloud application on its native architecture. The
purpose of the fog is to perform low latency calculations /
aggregations on the data while routing it to the central cloud for