778 IEEE TRANSACTIONS ON AUTOMATIC CONTROL, VOL. 55, NO. 3, MARCH 2010
Consensus of a Class of Second-Order Multi-Agent Systems
With Time-Delay and Jointly-Connected Topologies
Peng Lin and Yingmin Jia, Member, IEEE
Abstract—This technical note investigates consensus problems of a class
of second-order continuous-time multi-agent systems with time-delay and
jointly-connected topologies. We first introduce a neighbor-based linear
protocol with time-delay. Then we derive a sufficient condition in terms
of linear matrix inequalities (LMIs) for average consensus of the system.
Furthermore, we discuss the case where the time-delay affects only the in-
formation that is being transmitted and show that consensus can be reached
with arbitrary bounded time-delay. Finally, simulation results are provided
to demonstrate the effectiveness of our theoretical results.
Index Terms—Average consensus, jointly-connected topologies, second-
order dynamics, time-delay.
I. I
NTRODUCTION
Consensus problems of multi-agent systems have drawn substantial
research effort from many researchers [1]–[24], due to recent techno-
logical advances in communication and computation, and their impor-
tant practical applications such as cooperative control of unmanned
vehicles and control of communication networks. In the past decade,
many studies have been conducted on consensus problems for first-
order multi-agent systems [1]–[11]. In [1], Vicsek et al. proposed a
simple model for the phase transition of a group of self-driven particles
and demonstrated by simulation that the headings of all agents con-
verge to a common value. In [2], Jadbabaie et al. studied the observed
behavior reported in [1] and analyzed the alignment of an undirected
network of agents with switching topologies that are periodically con-
nected. Moreover, the work of [3] and [4] extended the results of [2]
and presented some more relaxable conditions for consensus of infor-
mation under dynamically changing topologies. By a Lyapunov-based
approach, Olfati-Saber et al. investigated a systematical framework of
consensus problems and gave conditions for average consensus of the
system with directed graphs or time-delay [5]. Recently, more atten-
tion has been paid to consensus problems for second-order multi-agent
systems [12]–[20]. For example, Ren et al. proposed a second-order
protocol and provided sufficient conditions for networks with fixed and
switching topologies [14]. Founded on the work of [14], Lin et al. gave
sufficient conditions for state consensus of the system with switching
topologies and time-delay on connected graphs [15]. Also, Hong et
al. studied a leader-follower scheme with jointly-connected topolo-
gies by a Lyapunov-based approach and related space decomposition
Manuscript received February 04, 2009; revised July 02, 2009, November
10, 2009, November 12, 2009, and November 16, 2009. First published
February 05, 2010; current version published March 10, 2010. This work
was supported by the NSFC (60727002,60774003,90916024,60921001), the
COSTIND (A2120061303), and the National 973 Program (2005CB321902).
Recommended by Associate Editor Z. Qu.
P. Lin is with the Seventh Research Division and the Department of Systems
and Control, Beihang University, Beijing 100191, China. He is also with the
Institute of Astronautics and Aeronautics, University of Electronic Science and
Technology of China, Chengdu 610054, China (e-mail: lin_peng0103@sohu.
com).
Y. Jia is with the Seventh Research Division and the Department of Systems
and Control, Beihang University, Beijing 100191, China. He is also with the
Key Laboratory of Mathematics Informatics and Behavioral Semantics, Bei-
hang University, Beijing 100191, China (e-mail: ymjia@buaa.edu.cn).
Digital Object Identifier 10.1109/TAC.2010.2040500
technique [16]. Liu et al. addressed an asynchronous discrete-time for-
mulation with fixed topology and derived conditions under which a
multi-agent system achieves cohesiveness in the presence of time-delay
and uncertainty [17]. However, most of the existing results for second-
order multi-agent systems are concerned with the networks with zero
time-delay on jointly-connected topologies or nonzero time-delay on
connected topologies, and there are few results available to treat the
networks with jointly-connected topologies and time-delay.
In this technical note, we study consensus problems of second-order
continuous-time multi-agent systems in a way to extend the results
in [5]. The communication topology considered is jointly-connected
and coupled with time-delay, different from [15], where each possible
communication topology is required to be connected. As commonly
known, it is much harder to study consensus problems on jointly-con-
nected topologies than on connected topologies, especially when time-
delays are involved. Based on the local velocity information and the
distributed relative position information, we introduce a delayed linear
consensus protocol, and by employing a contradiction approach, we
derive a sufficient condition in terms of LMIs for average consensus of
the system. The condition is composed of a set of LMIs and each LMI
corresponds to one possible connected component of the communica-
tion topology, making it possible to simplify the network analysis just
by studying connected components with different topology structures.
In addition, we also discuss the case where the time-delay affects only
the information that is being transmitted, and show that consensus can
still be reached with arbitrary bounded time-delay.
The following notations will be used throughout this technical
note.
m
denotes the set of all
m
dimensional real column vec-
tors;
I
m
denotes the
m
dimensional unit matrix;
x
i
denotes the
i
th
component of the vector
x
;
denotes the kronecker product;
1
represents
[1
;
1
;
111
;
1]
T
with compatible dimensions (sometimes, we
use
1
n
to denote
1
with dimension
n
); 0 denotes a zero value or a
zero matrix with appropriate dimensions;
k1k
refers to the standard
Euclidean norm of vectors; the symbol
3
denotes the symmetric term
of a symmetric matrix;
C
([
a; b
]
;
n
)
represents the Banach space of
continuous functions mapping the interval
[
a; b
]
into
n
.
II. G
RAPH THEORY AND
CONSENSUS PROTOCOL
A. Graph Theory
In this subsection, we present some definitions and results in
graph theory that will be used in this technical note (referring to
[25]). Let
G
(
V
;
E
;
A
)
be an undirected graph of order
n
, where
V
=
f
s
1
;
111
;
s
n
g
is the set of nodes,
EV2V
is the set of edges,
and
A
=[
a
ij
]
is a weighted adjacency matrix. The node indexes
belong to a finite set
I
=
f
1
;
2
;
111
;n
g
. An edge of
G
is denoted by
e
ij
=(s
i
;
s
j
)
. Since the graph considered is undirected, it means that
once
e
ij
2E
, then
e
ji
2E
. The adjacency matrix is defined as
a
ii
=0
and
a
ij
=
a
ji
0
, where
a
ij
>
0
if and only if
e
ij
2E
. The set of
neighbors of node
s
i
is denoted by
N
i
=
f
s
j
2V
:(s
i
;
s
j
)
2Eg
.
The Laplacian corresponding to the undirected graph is defined as
L
=[
l
ij
]
, where
l
ii
=
n
j
=1
a
ij
and
l
ij
=
0
a
ij
,
i
6
=
j
. Obviously,
L
=
L
T
for any undirected graph. A path is a sequence of ordered
edges of the form
(s
i
;
s
i
)
;
(s
i
;
s
i
)
;
111
, where
s
i
2V
. If there is
a path from every node to every other node, the graph is said to be
connected. The union of a collection of graphs
G
1
;
111
;
G
m
, with the
same node set
V
, is defined as the graph
G
1
0
m
, whose node set is
V
and edge set is the union of the edge sets of all graphs in the collection.
Moreover, this collection,
G
1
;
111
;
G
m
is jointly-connected if its
union graph
G
1
0
m
is connected.
0018-9286/$26.00 © 2010 IEEE