1551-3203 (c) 2019 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.2019.2927292, IEEE
Transactions on Industrial Informatics
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS, VOL. XX, NO. Y, MONTH 201X 1
Rapid-flooding Time Synchronization
for Large-scale Wireless Sensor Networks
Fanrong Shi, Student Member, IEEE, Xianguo Tuo, Simon X. Yang, Senior Member, IEEE ,
Jing Lu and Huailiang Li, Member, IEEE
Abstract—Accurate and fast-convergent time synchronization
is very important for wireless sensor networks. The flooding
time synchronization converges fast, but its transmission delay
and by-hop error accumulation seriously reduce the synchro-
nization accuracy. In this paper, a rapid-flooding multiple one-
way broadcast time-synchronization (RMTS) protocol for large-
scale wireless sensor networks is proposed. To minimize the by-
hop error accumulation, the RMTS uses maximum likelihood
estimations for clock skew estimation and clock offset estimation,
and quickly shares the estimations among the networks. As a
result, the synchronization error resulting from delays is greatly
reduced, while faster convergence and higher-accuracy synchro-
nization is achieved. Extensive experimental results demonstrate
that, even over 24-hops networks, the RMTS is able to build
accurate synchronization at the third synchronization period,
and moreover, the by-hop error accumulation is slower when
the network diameter increases.
Index Terms—Rapid-flooding time synchronization, maximum
likelihood estimation, one-way broadcast, fast convergence.
I. INTRODUCTION
T
IME synchronization is very important to wireless sensor
network (WSN) applications, e.g., data acquisition [1],
[2], low power [3], [4], location services [5], [6], security
[7], [8], networked control [9], [10], industrial WSNs [11],
[12], and smart grid measurement [13], [14]. The traditional
network time synchronization protocols, e.g., network time
protocol and global positioning system (GPS), may not meet
the synchronization requirements in energy-constrained WSN
applications resulting from the extra hardware needed or com-
plex protocol employed [15]. The aim of time synchronization
algorithms is to correct the local time information on nodes
and drive the entire network to obtain a time notion of
consistent values.
Manuscript received September 25, 2018; revised October 16 and May
19, 2019; accepted June 25, 2019. Date of publication XX XX, XXXX;
date of current version XX X, XX. This work was supported in part
by National Natural Science Foundation of China Programs (61601383),
Sichuan Science and Technology Program (No.2018GZ0095), and Longshan
academic talent research supporting program of Southwest University of
Science and Technology (17LZX650,18LZX633). (Corresponding authors:
Simon X. Yang, Xianguo Tuo, Huailiang Li.)
Fanrong Shi and Jing Lu are with School of Information Engineering,
Southwest University of Science and Technology, Mianyang 621010, China
(e-mail: sfr
swust@swust.edu.cn; lujing 017@live.cn).
Xianguo Tuo is with Sichuan University of Science and Engineering,
Zigong 643000, China (e-mail:tuoxg@cdut.edu.cn).
Simon X. Yang is with Advanced Robotics and Intelligent Systems (ARIS)
Laboratory, School of Engineering, University of Guelph, Guelph, Ontario,
N1G 2W1, Canada (email: syang@uoguelph.ca).
Huailiang Li is with the College of Geophysics, Chengdu University of
Technology, Chengdu 610059, China. (email: lihl@cdut.edu.cn).
Low synchronization error, rapid synchronization conver-
gence, and weak-dependency topology management are very
important requirements of robust time synchronization in
large-scale WSN applications. A faster-convergence approach
may adapt rapidly to the changes in clock drifts and network
topology, and recover quickly from loss of synchronization.
It is difficult to meet all of the above requirements due
to the transmission delay, topology changes, and clock drifts.
The RBS [16] and CESP [17] broadcast periodically to build
accurate time synchronization among all of the receivers,
while fail to meet the synchronization requirements over
larger distances [18]. The TPSN [19] is more accurate than
the RBS due to less clock offset estimation error on two-
way message exchange models, but it is not a distributed
approach as topology management is needed to maintain a
spanning tree. Average-consensus-based protocols, e.g., GTSP
[20], ATS [21], CCS [22], DISTY [23], and DiStiNCT [24],
are completely independent of topology and more robust to
a dynamic WSN, but they cost many more synchronization
periods to build the time synchronization. For instance, the
synchronization convergence time is up to 120 rounds of
synchronization periods in a 5 × 7 grid (diameter of 10) for
ATS [21]. At present, there is no good method to shorten
the convergence time for these approaches. The maximum-
consensus-based protocols MTS [25] and SMTS [26] converge
faster than ATS, but they still cost approximately 90 rounds
to converge in a 20-node ring network (diameter of 10), and
their convergence time is linear growth of the diameter [25].
The flooding time synchronization protocols, e.g., FTSP
[27], EGSync [28], Glossy [29], PulseSync [18], FCSA
[30], PISync (FloodPISync and PulsePISync) [31], attracted
our attention because they have the potential to be a fast-
convergence, accurate, and distributed time synchronization
algorithm. The FTSP, FCSA and FloodPISync are slow-
flooding protocols, in which the nodes broadcast their local
time information periodically and asynchronously, and all of
the network nodes synchronize with the root node (reference)
when the synchronization is convergent. However, the time
information on the root cannot be flooded with the multiple
hop nodes quickly, and the accuracy of the time information is
reduced by the clock drift on the flooding path. Additionally,
the FCSA must maintain the neighbor node table. A rapid-
flooding time synchronization protocol (e.g., PulsePISync,
PulseSync or Glossy), in which the time information of the
root is forwarded to every node as fast as possible, can adapt
quickly to changes in clock drift. If the time information
is flooded with nodes in a very short time, then there is
剩余9页未读，继续阅读
资源评论
