Liao et al.

Advances in Difference Equations (2019) 2019:123

https://doi.org/10.1186/s13662-019-2006-7

RESEARCH Open Access

A new result on stability analysis for discrete

system with interval time-varying delays
Daixi Liao1,2* , Shouming Zhong1 , Jun Cheng3* , Can Zhao1 , Xiaojun Zhang1 and Yongbin Yu4
*
Correspondence:
liaodaixizaici@163.com; Abstract
jcheng6819@126.com
1
School of Mathematical Sciences, This paper is concerned with the stability criteria for a discrete-time linear system with
University of Electronic Science and interval time-varying delays. By using the time delay division we construct an
Technology of China, Chengdu, augmented Lyapunov–Krasovskii functional for two delay subintervals. Moreover, we
P.R. China
3
School of Automation and use a new summation inequality to estimate the derivatives of LKFs more accurately
Electronic Engineering, Qingdao and derive less conservative criteria. Finally, we present two numerical examples to
University of Science and demonstrate that the obtained results are less conservative.
Technology, Qingdao, P.R. China
Full list of author information is Keywords: Discrete system; Stability; Lyapunov–Krasovskii functional; Interval
available at the end of the article
time-varying delay

1 Introduction
During the transmission and processing of signal or data, time delay inevitably occurs in
such practical systems as neural networks and networked control systems. It is well known
that the presence of time delay often affects the dynamic behavior of the system, resulting
in deterioration or even instability in system performance. Therefore, the research on the
stability of time delay systems has been the focus of many scholars [1–5].
One essential research focus on a system with time delay is how to reduce the conser-
vatism of the stability condition. In view of this, many researchers choose to approach this
from two perspectives, selecting an appropriate Lyapunov–Krasovskii functional (LKF)
and making a more accurate estimation of its derivative. With regard to the former aspect,
many significant results have been achieved such as constructing an amplified LKF [6], in-
creasing its integral multiplicity [7], or dividing the time delay interval [8]. With regard to
the latter, the Jensen inequality [9, 10] is the first powerful tool used to estimate the deriva-
tive of LKF in the stability theory of time-delay systems. Later on, the Wirtinger-based
integral inequality [11, 12], the free-matrix-based integral inequality [13–15], and a series
of integral inequalities [16] have been proposed. These inequalities are widely applied in
time delay systems [17–28]. Of course, there are other methods or combinations of differ-
ent methods, such as the free-weighting matrix approach proposed in the literature [29],
the basic idea is introducing some free weighting matrix through the Newton–Leibniz for-
mula and/or the equation of state of the system, and the less conservative stability results
can also be obtained.
It is worth noting that not only continuous-time systems, but also discrete-time systems
play an important role and have received extensive attention in the literature [30–47]. In

© The Author(s) 2019. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License
(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, pro-
vided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and
indicate if changes were made.
Liao et al. Advances in Difference Equations (2019) 2019:123 Page 2 of 12

[12], by using the Wirtinger-based inequality an improved stability condition is obtained.

However, the allowable delay upper bound can be further lifted. This inspires our research
in this paper.
Based on the above discussions, by means of a new summation inequality and time delay
division, less conservative criteria than some existing ones are obtained, and two examples
are presented to demonstrate the effectiveness and superiority of the proposed approach.
The main contributions of this paper are as follows:
(i) A new summation inequality is obtained by the Schmidt orthogonalization method,
and the corresponding result is less conservative.
(ii) The proof of Corollary 1 in [15] can be simplified by the substitution of variables,
which is demonstrated by Lemma 2.1 in this paper.
Notation: Throughout this paper, n denotes the n-dimensional Euclidean space, and
n×m is the set of all n × m real matrices; I is used to denote the identity matrix of proper
dimension; diag{· · ·} denotes block diagonal matrix, and f , g denotes the inner product
of f and g; P > 0 (≥ 0) means that P is a real symmetric and positive definite (semipositive
definite) matrix; the symmetric term in a matrix is denoted by ∗, and sym[A] = A + AT ;
N + (N) is the set of all positive (nonnegative) integers.

2 Preliminaries
Consider the following discrete time-varying delay system:
⎧
⎨x(k + 1) = Ax(k) + Bx(k – d(k)), k ∈ N,
(1)
⎩x(k) = φ(k), k = –d2 , –d2 + 1, . . . , 0,

where x(k) = (x1 (k), x2 (k), . . . , xn (k))T ∈ n is the state vector, φ(k) ∈ n is the initial condi-
tion, n ∈ N + , A and B ∈ n×n are constant matrices, and d(k) is the discrete time-varying
delay satisfying d1 ≤ d(k) ≤ d2 , where d1 and d2 are nonnegative integers.

Lemma 2.1 Let x(i) be a sequence of discrete-time variables in [a, a+n] → n . For symmet-
ric matrices R ∈ n×n and Z11 , Z22 , Z33 ∈ 4n×4n and for any matrices Z12 , Z13 , Z23 ∈ 4n×4n
and N1 , N2 , N3 ∈ 4n×n satisfying

⎛ ⎞
Z11 Z12 Z13 N1
⎜ ∗ N2 ⎟
⎜ Z22 Z23 ⎟
Ξ =⎜ ⎟ ≥ 0,
⎝ ∗ ∗ Z33 N3 ⎠
∗ ∗ ∗ R

we have the following inequality:

a+n–1
– ηT (i)Rη(i) ≤ ωT Ωω, (2)
i=a

where

η(i) = x(i + 1) – x(i),

Liao et al. Advances in Difference Equations (2019) 2019:123 Page 3 of 12

a+n a+n a+n T

T 1 T T 2 T
ω = x (a + n), x (a), x (i), x (j) ,
n+1 i=a
(n + 1)(n + 2) i=a j=i
 
1 1
Ω = n Z11 + Z22 + Z33 + sym[N1 Π1 + N2 Π2 + N3 Π3 ],
3 5
Π1 = ε1 – ε2 , Π2 = ε1 + ε2 – 2ε3 , Π3 = ε1 – ε2 + 6ε3 – 6ε4 ,
   
ε1 = I 0 0 0 , ε2 = 0 I 0 0 ,
   
ε3 = 0 0 I 0 , ε4 = 0 0 0 I .

Proof Define

f0 (i) = 1, f1 (i) = i, f2 (i) = i2 .

n
By the Schmidt orthogonalization method, where f , g = i=1 f (i)g(i), we obtain the fol-
lowing orthogonal polynomials gk (i), k = 0, 1, 2:

n+1 (n + 1)(n + 2)
g0 (i) = 1, g1 (i) = i – , g2 (i) = i2 – (n + 1)i + .
2 6

Then converting the coefficient of constant term to 1 or –1 by multiplying by a positive

number, we have

2 6 6
h0 (i) = 1, h1 (i) = –1 + i, h2 (i) = 1 – i+ i2 ,
n+1 n+2 (n + 1)(n + 2)

and hk (i), k = 0, 1, 2, satisfy

n–1 (n – 1)(n – 2)
h0 , h0  = n, h1 , h1  = n , h2 , h2  = n .
3(n + 1) 5(n + 1)(n + 2)

1 n+1 T
Let ς(i) = (h0 (i)uT , h1 (i)uT , h2 (i)uT , ξ T (i))T , where u = (yT (n + 1), yT (1), n+1 i=1 y (i),
 n+1  n+1 T
j=i y (j)) , y(i) is a sequence of discrete-time variable in [1, 1 + n] →  ,
2 T n
(n+1)(n+2) i=1
n+1 n+1 n+1
and ξ (i) = y(i + 1) – y(i). Noticing that i=1 iy(i) = i=1 j=i y(j), we can obtain

n
hk (i)ξ (i) = Πk+1 u, k = 0, 1, 2.
i=1

It is easy to see that ς T (i)Ξ ς(i) ≥ 0. Then summing both sides of the inequality before
from 1 to n yields

n
– ξ T (i)Rξ (i) ≤ uT Ω̃u,
i=1

n–1 (n–1)(n–2)
where Ω̃ = n(Z11 + 3(n+1) Z22 + 5(n+1)(n+2) Z33 ) + sym[N1 Π1 + N2 Π2 + N3 Π3 ].
Let x(i) = y(i – a + 1) be a sequence of discrete-time variables in [a, a + n] → n satisfying

a+n–1 n
– ηT (i)Rη(i) = – ξ T (i)Rξ (i) ≤ uT Ω̃u = ωT Ω̃ω.
i=a i=1
Liao et al. Advances in Difference Equations (2019) 2019:123 Page 4 of 12

Table 1 Allowable delay upper bound d2 for various d1 for Example 1

Method d1 2 4 6 10 15 20 25 30
[32] d2 13 13 14 15 18 22 26 30
[33] d2 13 13 14 17 20 24 29 33
[20] d2 14 15 16 18 21 25 30 34
[21] d2 17 17 18 20 23 27 31 35
[34] d2 18 18 19 20 23 26 30 35
[40] d2 20 21 21 22 24 27 29 34
[43] (Theorem 2) d2 22 22 22 23 25 28 32 36
[12] (Remark 4) d2 20 21 21 23 25 29 32 36
[12] (Theorem 1) d2 22 22 22 23 26 29 32 36
Remark 3 d2 21 22 22 23 25 29 32 36
Theorem 3.1 d2 22 22 22 23 26 29 32 36

Table 2 Allowable delay upper bound d2 for various d1 for Example 2

Method d1 2 5 7 10 20
[33] d2 7 9 11 14 24
[21] d2 9 11 13 16 26
[40] d2 13 14 16 19 29
[12] (Theorem 1) d2 14 17 18 21 31
Theorem 3.1 d2 15 17 18 21 31

2n 6n 2
Define Ω = Ω̃ + 3(n+1) Z22 + 5(n+1)(n+2) Z33 . Then Ω is affinely dependent on n, and we obtain
(2). This completes the proof. 

12 30 20
Remark 1 Define f3 (i) = i3 . Then h3 (i) = –1 + λ n+3 i– (n+2)(n+3)
i2 + (n+1)(n+2)(n+3)
i3 , where
6n2 +15n+11
λ= ,
and it is not the fourth Legendre polynomial anymore [16]. Thus deriving
6(n+1)(n+2)
a general form of the discrete inequality (2) is still an interesting and challenging task.

Remark 2 By the substitution of variables demonstrated in Lemma 2.1, the proof of Corol-
lary 1 in [15] can be simplified. By means of the new summation inequality the results cal-
culated based on the criteria given in this paper are less conservative than those reported
in the existing literature from Tables 1 and 2.

3 Main results
For reducing the conservatism, we further present a delay partitioning method, where the
time delay interval is divided into two segments: [d1 , d2 ] = [d1 , d] ∪ [d, d2 ].

Theorem 3.1 For given positive integers d1 and d2 satisfying d1 ≤ d(k) ≤ d2 , system (1) is
asymptotically stable if there exist matrices P > 0 (∈ 4n×4n ), Qi > 0 (∈ n×n ), symmetric
matrices R ∈ n×n and Zj11 , Zj22 , Zj33 ∈ 4n×4n , any matrices Zj12 , Zj13 , Zj23 ∈ 4n×4n and
Nji ∈ 4n×n (i = 1, 2, 3; j = 1, 2, . . . , 7) such that the following LMIs hold:

       
Ψ1 d(k) = D1 + D3 + D4 + D5 + D6 + G1 d(k) + G3 d(k) + G4 d(k) + H1 + H2

< 0, d(k) ∈ [d1 , d] ∩ N, (3)

       
Ψ2 d(k) = D2 + D3 + D4 + D7 + D8 + G2 d(k) + G5 d(k) + G6 d(k) + H1 + H3

< 0, d(k) ∈ [d, d2 ] ∩ N, (4)

Liao et al. Advances in Difference Equations (2019) 2019:123 Page 5 of 12

⎛ ⎞ ⎧
Zj11 Zj12 Zj13 Nj1 ⎪
⎜ ∗ ⎪
⎨1 if j = 1,
⎜ Zj22 Zj23 Nj2 ⎟
⎟
Ξj = ⎜ ⎟ ≥ 0, and i = 2 if j = 2, 3, 5, (5)
⎝ ∗ ∗ Zj33 Nj3 ⎠ ⎪
⎪
⎩
∗ ∗ ∗ Ri 3 if j = 4, 6, 7,

where

D1 = E2T PE2 – E1T PE1 ,

D2 = E4T PE4 – E3T PE3 ,

D3 = eT1 Q1 e1 – eT2 (Q1 – Q2 )e2 – eT4 (Q2 – Q3 )e4 – eT5 Q3 e5 ,

D4 = d1 (e0 – e1 )T R1 (e0 – e1 ) + α1 sym[N11 Π1 + N12 Π2 + N13 Π3 ]α1T ,

D5 = (d – d1 )(e0 – e1 )T R2 (e0 – e1 ) + α2 sym[N21 Π1 + N22 Π2 + N23 Π3 ]α2T

+ α3 sym[N31 Π1 + N32 Π2 + N33 Π3 ]α3T ,

D6 = (d2 – d)(e0 – e1 )T R3 (e0 – e1 ) + α4 sym[N41 Π1 + N42 Π2 + N43 Π3 ]α4T ,

D7 = (d – d1 )(e0 – e1 )T R2 (e0 – e1 ) + α5 sym[N51 Π1 + N52 Π2 + N53 Π3 ]α5T ,

D8 = (d2 – d)(e0 – e1 )T R3 (e0 – e1 ) + α6 sym[N61 Π1 + N62 Π2 + N63 Π3 ]α6T

+ α7 sym[N71 Π1 + N72 Π2 + N73 Π3 ]α7T ,

    
G1 d(k) = sym (E2 – E1 )T PF1 d(k) ,
    
G2 d(k) = sym (E4 – E3 )T PF2 d(k) ,
  
    1 1
G3 d(k) = α2 d(k) – d1 Z211 + Z222 + Z233 α2T ,
3 5
  
    1 1
G4 d(k) = α3 d – d(k) Z311 + Z322 + Z333 α3T ,
3 5
  
    1 1
G5 d(k) = α6 d(k) – d Z611 + Z622 + Z633 α6T ,
3 5
  
    1 1
G6 d(k) = α7 d2 – d(k) Z711 + Z722 + Z733 α7T ,
3 5
     T   T
F1 d(k) = 0, 0, d(k) – d1 + 1 e7 + d – d(k) + 1 eT8 , 0 ,
       T
F2 d(k) = 0, 0, d(k) – d + 1 eT8 + d2 – d(k) + 1 eT9 , 0 ,

E1 = eT1 , (d1 + 1)eT6 – eT1 , (d2 – d + 1)eT9 – eT2 – eT3 – eT4 ,
T
1 T T
(d1 + 1)(d1 + 2)e10 – (d1 + 1)e1 ,
2

E2 = eT0 , (d1 + 1)eT6 – eT2 , (d2 – d + 1)eT9 – eT3 – eT4 – eT5 ,
T
1
(d1 + 1)(d1 + 2)eT10 – (d1 + 1)eT6 ,
2
Liao et al. Advances in Difference Equations (2019) 2019:123 Page 6 of 12


E3 = eT1 , (d1 + 1)eT6 – eT1 , (d – d1 + 1)eT7 – eT2 – eT3 – eT4 ,
T
1 T T
(d1 + 1)(d1 + 2)e10 – (d1 + 1)e1 ,
2

E4 = eT0 , (d1 + 1)eT6 – eT2 , (d – d1 + 1)eT7 – eT3 – eT4 – eT5 ,
T
1
(d1 + 1)(d1 + 2)eT10 – (d1 + 1)eT6 ,
2
  
1 1
H1 = α1 d1 Z111 + Z122 + Z133 α1T ,
3 5
  
1 1
H2 = α4 (d2 – d) Z411 + Z422 + Z433 α4T ,
3 5
  
1 1
H3 = α5 (d – d1 ) Z511 + Z522 + Z533 α5T ,
3 5

and
     
α1 = eT1 , eT2 , eT6 , eT10 , α2 = eT2 , eT3 , eT7 , eT11 , α3 = eT3 , eT4 , eT8 , eT12 ,
     
α4 = eT4 , eT5 , eT9 , eT13 , α5 = eT2 , eT4 , eT7 , eT11 , α6 = eT4 , eT3 , eT8 , eT12 ,
 
α7 = eT3 , eT5 , eT9 , eT13 , ei = (0n×(i–1)n , I, 0n×(13–i)n ), i = 1, 2, . . . , 13,

e0 = Ae1 + Be3 .

Proof Construct a Lyapunov functional as follows:

5
V (k) = Vi (k),
i=1

where

V1 (k) = θ T (k)Pθ (k),

k–1 k–d1 –1 k–d–1
V2 (k) = xT (i)Q1 x(i) + xT (i)Q2 x(i) + xT (i)Q3 x(i),
i=k–d1 i=k–d i=k–d2

–1 k–1
V3 (k) = ηT (j)R1 η(j),
i=–d1 j=k+i

–d1 –1 k–1
V4 (k) = ηT (j)R2 η(j),
i=–d j=k+i

–d–1 k–1
V5 (k) = ηT (j)R3 η(j),
i=–d2 j=k+i

and
k–1 k–d1 –1 k–1 k–1 T
T T T T
θ (k) = x (k), x (i), x (i), x (j) .
i=k–d1 i=k–d2 i=k–d1 j=i
Liao et al. Advances in Difference Equations (2019) 2019:123 Page 7 of 12

For simplicity, we first define the following vectors:

  
ξ1T (k) = xT (k) xT (k – d1 ) xT k – d(k) xT (k – d) xT (k – d2 )

U1T U2T U3T U4T W1T W2T W3T W4T ,
  
ξ2T (k) = xT (k) xT (k – d1 ) xT k – d(k) xT (k – d) xT (k – d2 )

U1T U5T U6T U7T W1T W5T W6T W7T ,

where
k
1
U1 = x(i),
d1 + 1
i=k–d1
k–d1
1
U2 = x(i),
d(k) – d1 + 1
i=k–d(k)
k–d(k)
1
U3 = x(i),
d – d(k) + 1
i=k–d
k–d
1
U4 = x(i),
d2 – d + 1
i=k–d2
k–d1
1
U5 = x(i),
d – d1 + 1
i=k–d
k–d
1
U6 = x(i),
d(k) – d + 1
i=k–d(k)
k–d(k)
1
U7 = x(i),
d2 – d(k) + 1
i=k–d2
k k
2
W1 = x(j),
(d1 + 1)(d1 + 2)
i=k–d1 j=i
k–d1 k–d1
2
W2 = x(j),
(d(k) – d1 + 1)(d(k) – d1 + 2)
i=k–d(k) j=i
k–d(k) k–d(k)
2
W3 = x(j),
(d – d(k) + 1)(d – d(k) + 2) j=i
i=k–d
k–d k–d
2
W4 = x(j),
(d2 – d + 1)(d2 – d + 2)
i=k–d2 j=i
k–d1 k–d1
2
W5 = x(j),
(d – d1 + 1)(d – d1 + 2)
i=k–d j=i
k–d k–d
2
W6 = x(j),
(d(k) – d + 1)(d(k) – d + 2)
i=k–d(k) j=i
k–d(k) k–d(k)
2
W7 = x(j).
(d2 – d(k) + 1)(d2 – d(k) + 2) j=i
i=k–d2
Liao et al. Advances in Difference Equations (2019) 2019:123 Page 8 of 12

Case (i): d(k) ∈ [d1 , d] ∩ N . The forward difference of Vi (k) (i = 1, 2, . . . , 5) is given by

V1 (k) = θ T (k + 1)Pθ (k + 1) – θ T (k)Pθ (k)

   
= ξ1T (k) E2T PE2 – E1T PE1 + sym (E2 – E1 )T PF1 d(k) ξ1 (k)
  
= ξ1T (k) D1 + G1 d(k) ξ1 (k), (6)

V2 (k) = xT (k)Q1 x(k) – xT (k – d1 )Q1 x(k – d1 ) + xT (k – d1 )Q2 x(k – d1 )

– xT (k – d)Q2 x(k – d) + xT (k – d)Q3 x(k – d) – xT (k – d2 )Q3 x(k – d2 )

 
= ξiT (k) eT1 Q1 e1 – eT2 (Q1 – Q2 )e2 – eT4 (Q2 – Q3 )e4 – eT5 Q3 e5 ξi (k)

= ξiT (k)D3 ξi (k), i = 1, 2. (7)

Using Lemma 2.1, we can obtain

–1 k –1 k–1
V3 (k) = ηT (j)R1 η(j) – ηT (j)R1 η(j)
i=–d1 j=k+i+1 i=–d1 j=k+i

k–1
= d1 ηT (k)R1 η(k) – ηT (i)R1 η(i)
i=k–d1
 
≤ ξiT (k) d1 (e0 – e1 )T R1 (e0 – e1 ) ξi (k)
  
T
 T T T T 1 1
+ ξi (k) e1 , e2 , e6 , e10 d1 Z111 + Z122 + Z133
3 5

 T
+ sym[N11 Π1 + N12 Π2 + N13 Π3 ] eT1 , eT2 , eT6 , eT10 ξi (k)

= ξiT (k)[D4 + H1 ]ξi (k), i = 1, 2, (8)

–d1 –1 k –d1 –1 k–1
V4 (k) = ηT (j)R2 η(j) – ηT (j)R2 η(j)
i=–d j=k+i+1 i=–d j=k+i

k–d1 –1 k–d(k)–1
= (d – d1 )ηT (k)R2 η(k) – ηT (i)R2 η(i) – ηT (i)R2 η(i)
i=k–d(k) i=k–d

     
≤ ξ1T (k) (d – d1 )(e0 – e1 )T R2 (e0 – e1 ) ξ1 (k) + ξ1T (k) eT2 , eT3 , eT7 , eT11 d(k) – d1
  
1 1
× Z211 + Z222 + Z233 + sym[N21 Π1 + N22 Π2 + N23 Π3 ]
3 5
 T T T T T
× e2 , e3 , e7 , e11 ξ1 (k)
  
T
 T T T T   1 1
+ ξ1 (k) e3 , e4 , e8 , e12 d – d(k) Z311 + Z322 + Z333
3 5

 T
+ sym[N31 Π1 + N32 Π2 + N33 Π3 ] eT3 , eT4 , eT8 , eT12 ξ1 (k)
    
= ξ1T (k) D5 + G3 d(k) + G4 d(k) ξ1 (k), (9)
Liao et al. Advances in Difference Equations (2019) 2019:123 Page 9 of 12

–d–1 k –d–1 k–1

V5 (k) = ηT (j)R3 η(j) – ηT (j)R3 η(j)
i=–d2 j=k+i+1 i=–d2 j=k+i

   T T T T
≤ ξ1T (k) T T
(d2 – d)(e0 – e1 ) R3 (e0 – e1 ) ξ1 (k) + ξ1 (k) e4 , e5 , e9 , e13 (d2 – d)
  
1 1
× Z411 + Z422 + Z433 + sym[N41 Π1 + N42 Π2 + N43 Π3 ]
3 5
 T T T T T
× e4 , e5 , e9 , e13 ξ1 (k)

= ξ1T (k)[D6 + H2 ]ξ1 (k). (10)

From (6)–(10) we have V (k) ≤ ξ1T (k)Ψ1 (d(k))ξ1 (k). According to (3), we can obtain
V (k) < 0 for any ξ1 (k) = 0 and d(k) ∈ [d1 , d] ∩ N .
Case (ii): d(k) ∈ [d, d2 ] ∩ N . Similarly, we have

V1 (k) = θ T (k + 1)Pθ (k + 1) – θ T (k)Pθ (k)

   
= ξ2T (k) E4T PE4 – E3T PE3 + sym (E4 – E3 )T PF2 d(k) ξ2 (k)
  
= ξ2T (k) D2 + G2 d(k) ξ2 (k), (11)
–d1 –1 k –d1 –1 k–1
V4 (k) = ηT (j)R2 η(j) – ηT (j)R2 η(j)
i=–d j=k+i+1 i=–d j=k+i

   T T T T
≤ ξ2T (k) T T
(d – d1 )(e0 – e1 ) R2 (e0 – e1 ) ξ2 (k) + ξ2 (k) e2 , e4 , e7 , e11 (d – d1 )
  
1 1
× Z511 + Z522 + Z533 + sym[N51 Π1 + N52 Π2 + N53 Π3 ]
3 5
 T T T T T
× e2 , e4 , e7 , e11 ξ2 (k)

= ξ2T (k)[D7 + H3 ]ξ2 (k), (12)

–d–1 k –d–1 k–1
V5 (k) = ηT (j)R3 η(j) – ηT (j)R3 η(j)
i=–d2 j=k+i+1 i=–d2 j=k+i

k–d–1 k–d(k)–1
= (d2 – d)ηT (k)R3 η(k) – ηT (i)R3 η(i) – ηT (i)R3 η(i)
i=k–d(k) i=k–d2

     
≤ ξ2T (k) (d2 – d)(e0 – e1 )T R3 (e0 – e1 ) ξ2 (k) + ξ2T (k) eT4 , eT3 , eT8 , eT12 d(k) – d
  
1 1
× Z611 + Z622 + Z633 + sym[N61 Π1 + N62 Π2 + N63 Π3 ]
3 5
 T T T T T
× e4 , e3 , e8 , e12 ξ2 (k)
  
T
 T T T T   1 1
+ ξ2 (k) e3 , e5 , e9 , e13 d2 – d(k) Z711 + Z722 + Z733
3 5

 T
+ sym[N71 Π1 + N72 Π2 + N73 Π3 ] eT3 , eT5 , eT9 , eT13 ξ2 (k)
    
= ξ2T (k) D8 + G5 d(k) + G6 d(k) ξ2 (k), (13)
Liao et al. Advances in Difference Equations (2019) 2019:123 Page 10 of 12

and V (k) ≤ ξ2T (k)Ψ2 (d(k))ξ2 (k), which under condition (4) indicates that V (k) is neg-
ative definite.
Now by combining Case (i) and Case (ii), V (k) is negative definite, which guarantees
that system (1) is asymptotically stable. This completes the proof. 

Remark 3 To show the advantage of time delay division for the stability condition without
using time delay division, we omitted the proof process and only display the results in
Table 1.

4 Numerical illustrations
In this section, we present some numerical examples that often appear in the literature to
demonstrate the effectiveness and sophistication of the proposed approach.

Example 1 Consider system (1) with the following parameters:

0.8 0 –0.1 0
A= , B= .
0.05 0.9 –0.2 –0.1

Example 2 Consider system (1), where

0.7 0.1 –0.1 0.1

A= , B= .
0.05 0.7 –0.1 –0.2

The allowable delay upper bound d2 can be found for given d1 or vice versa. The simu-
lation results are listed in Tables 1 and 2.

Remark 4 In [12] and this paper, the stability conditions are obtained by using the
Wirtinger-based integral inequality and the free-matrix-based integral inequality, respec-
tively. We can see from Table 1 that the latter is superior to the former in reducing the
conservatism when time delay division is not used, whereas the conservatism of both is
the same when time delay division is used. Therefore, on the basis of some processing tech-
niques, such as the time delay division mentioned before, our future research is discussing
the relationship between the tightness of inequalities and the conservatism of criteria.

Remark 5 From Table 2 we can see that with the lower bounds d1 = 2, 5, 7, 10, 20, the cor-
responding d2 = 13, 14, 16, 19, 29 obtained in [40], d2 = 14, 17, 18, 21, 31 obtained by The-
orem 1 in [12], and d2 = 15, 17, 18, 21, 31 derived by Theorem 3.1 in this paper. Thus the
results obtained from Theorem 3.1 in this paper are significantly better than those ob-
tained from the existing methods.

5 Conclusion
In this paper, the problem of stability criteria of discrete systems with interval time-varying
delays has been investigated. By means of the discrete form of free-matrix-based integral
inequality and the time delay division, less conservative criteria than some existing ones
are derived due to a tighter estimation of the new inequality. Finally, some illustrative ex-
amples are presented to demonstrate the effectiveness of the obtained method.
Liao et al. Advances in Difference Equations (2019) 2019:123 Page 11 of 12

Acknowledgements
This work of D. Liao was supported in part by the National Natural Science Foundation of China under Grant 61771004,
Research Fund for International Young Scientists of National Natural Science Foundation of China under Grant
61550110248, Guangxi Natural Science Foundation Project (No. 2017GXNSFBA198179), Basic Ability Promotion Project for
Young and Middle-aged Teachers in Universities of Guangxi (No. 2018KY0214), and Sichuan Science and Technology Plan
Project (2018GZ0072, 2018JZ0065).

Funding
This research was partially supported by the National Natural Science Foundation of China under Grant 61771004,
Research Fund for International Young Scientists of National Natural Science Foundation of China under Grant
61550110248, Guangxi Natural Science Foundation Project (No. 2017GXNSFBA198179), Basic Ability Promotion Project for
Young and Middle-aged Teachers in Universities of Guangxi (No. 2018KY0214), and Sichuan Science and Technology Plan
Project (2018GZ0072, 2018JZ0065).

Availability of data and materials

Not applicable.

Competing interests
The authors declare that there is no conflict of interest regarding the publication of this paper.

Authors’ contributions
All authors drafted the manuscript, and they read and approved the final version.

Author details
1
School of Mathematical Sciences, University of Electronic Science and Technology of China, Chengdu, P.R. China.
2
School of Mathematical Sciences and Energy Engineering, Hunan Institute of Technology, Hengyang, P.R. China. 3 School
of Automation and Electronic Engineering, Qingdao University of Science and Technology, Qingdao, P.R. China. 4 School
of Information and Software Engineering, University of Electronic Science and Technology of China, Chengdu, P.R. China.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Received: 10 October 2018 Accepted: 4 February 2019

References
1. Zhang, D., Cheng, J., Park, J.H., Cao, J.: Robust H∞ control for nonhomogeneous Markovian jump systems subject to
quantized feedback and probabilistic measurements. J. Franklin Inst. 355(15), 6992–7010 (2018)
2. Liu, L., Meng, X.: Optimal harvesting control and dynamics of two-species stochastic model with delays. Adv. Differ.
Equ. 2017, 18 (2017)
3. Cheng, J., Chang, X.H., Park, J.H., Li, H., Wang, H.: Fuzzy-model-based H∞ control for discrete-time switched systems
with quantized feedback and unreliable links. Inf. Sci. 436–437, 181–196 (2018)
4. Cheng, J., Ahn, C.K., Karimi, H.R., Cao, J., Qi, W.: An event-based asynchronous approach to Markov jump systems with
hidden mode detections and missing measurements. IEEE Trans. Syst. Man Cybern. Syst. (2018, in press).
https://doi.org/10.1109/TSMC.2018.2866906
5. Cheng, J., Park, J.H., Cao, J., Zhang, D.: Quantized H∞ filtering for switched linear parameter-varying systems with
sojourn probabilities and unreliable communication channels. Inf. Sci. 466, 289–302 (2018)
6. Wu, M., Peng, C., Zhang, J., Fei, M., Tian, Y.: Further results on delay-dependent stability criteria of discrete systems
with an interval time-varying delay. J. Franklin Inst. 354, 4955–4965 (2017)
7. Xiong, L., Cheng, J., Cao, J., Liu, Z.: Novel inequality with application to improve the stability criterion for dynamical
systems with two additive time-varying delays. Appl. Math. Comput. 321, 672–688 (2018)
8. Li, Z., Bai, Y., Huang, C., Cai, Y.: Novel delay-partitioning stabilization approach for networked control system via
Wirtinger-based inequalities. ISA Trans. 61, 75–86 (2016)
9. Chen, J., Park, J.H., Xu, S.: Stability analysis of continuous-time systems with time-varying delay using new
Lyapunov–Krasovskii functionals. J. Franklin Inst. 355(13), 5957–5967 (2018)
10. Jiang, X., Han, Q., Yu, X.: Stability criteria for linear discrete-time systems with interval-like time-varying delay. In: Proc.
of the 2005, American Control Conference, pp. 2817–2822 (2005)
11. Seuret, A., Gouaisbaut, F.: Wirtinger-based integral inequality: application to time-delay systems. Automatica 49,
2860–2866 (2013)
12. Nam, P., Pathirana, P., Trinh, H.: Discrete Wirtinger-based inequality and its application. J. Franklin Inst. 352, 1893–1905
(2015)
13. Zeng, H., He, Y., Wu, M., She, J.: New results on stability analysis for systems with discrete distributed delay. Automatica
60, 189–192 (2015)
14. Zeng, H., He, Y., Wu, M., She, J.: Free-matrix-based integral inequality for stability analysis of systems with time-varying
delay. IEEE Trans. Autom. Control 60, 2768–2772 (2015)
15. Hua, C., Wu, S., Guan, X.: New robust stability condition for discrete-time recurrent neural networks with time-varying
delays and nonlinear perturbations. Neurocomputing 219, 203–209 (2017)
16. Chen, J., Xu, S., Zhang, B.: Single/multiple integral inequalities with applications to stability analysis of time-delay
systems. IEEE Trans. Autom. Control 62, 3488–3493 (2017)
17. Jiao, S., Shen, H., Wei, Y., Huang, X., Wang, Z.: Further results on dissipativity and stability analysis of Markov jump
generalized neural networks with time-varying interval delays. Appl. Math. Comput. 336, 338–350 (2018)
Liao et al. Advances in Difference Equations (2019) 2019:123 Page 12 of 12

18. Zhu, F., Meng, X., Zhang, T.: Optimal harvesting of a competitive n-species stochastic model with delayed diffusions.
Math. Biosci. Eng. 16, 1554–1574 (2019)
19. Luo, J., Tian, W., Zhong, S., Shi, K., Wang, W.: Non-fragile asynchronous event-triggered control for uncertain delayed
switched neural networks. Nonlinear Anal. Hybrid Syst. 29, 54–73 (2018)
20. Wang, B., Zhang, D., Cheng, J., Park, J.H.: Fuzzy model-based nonfragile control of switched discrete-time systems.
Nonlinear Dyn. 93(4), 2461–2471 (2018)
21. Liu, J., Zhang, J.: Note on stability of discrete-time time-varying delay systems. IET Control Theory Appl. 6(2), 335–339
(2012)
22. Shi, K., Tang, Y., Zhong, S., Yin, C., Huang, X., Wang, W.: Nonfragile asynchronous control for uncertain chaotic Lurie
network systems with Bernoulli stochastic process. Int. J. Robust Nonlinear Control 28, 1693–1714 (2018)
23. Lee, T., Park, J., Xu, S.: Relaxed conditions for stability of time-varying delay systems. Automatica 75, 11–15 (2017)
24. Wang, J., Shi, K., Huang, Q., Zhong, S., Zhang, D.: Stochastic switched sampled-data control for synchronization of
delayed chaotic neural networks with packet dropout. Appl. Math. Comput. 335, 211–230 (2018)
25. Zhang, C., He, Y., Jiang, L., Wu, M., Zeng, H.: Summation inequalities to bounded real lemmas of discrete-time systems
with time-varying delay. IEEE Trans. Autom. Control 62(5), 2582–2588 (2017)
26. Zhu, Q.: A dual internal model based repetitive control for linear discrete-time systems. IEEE Trans. Autom. Control
(2018, in press). https://doi.org/10.1109/TAC.2018.2836023
27. Zeng, H., Teo, K., He, Y., Wang, W.: Sampled-data-based dissipative control of T–S fuzzy systems. Appl. Math. Model. 65,
415–427 (2019)
28. Shi, K., Wang, J., Zhong, S., Zhang, X., Liu, Y., Cheng, J.: New reliable nonuniform sampling control for uncertain
chaotic neural networks under Markov switching topologies. Appl. Math. Comput. 347, 169–193 (2019)
29. Wu, M., He, Y., She, J., Liu, G.: Delay-dependent criteria for robust stability of time-varying delay systems. Automatica
40(8), 1435–1439 (2004)
30. Feng, T., Meng, X., Liu, L., Gao, S.: Application of inequalities technique to dynamics analysis of a stochastic
eco-epidemiology model. J. Inequal. Appl. 2016, 327 (2016)
31. Long, S., Zhong, S.: Improved results for stochastic stabilization of a class of discrete-time singular Markovian jump
systems with time-varying delay. Nonlinear Anal. Hybrid Syst. 23, 11–26 (2017)
32. Gao, H., Chen, T.: New results on stability of discrete-time systems with time-varying state delay. IEEE Trans. Autom.
Control 52(2), 328–334 (2007)
33. Zhang, B., Xu, S., Zou, Y.: Improved stability criterion and its applications in delayed controller design for discrete-
time systems. Automatica 44(11), 2963–2967 (2008)
34. Peng, C.: Improved delay-dependent stabilisation criteria for discrete systems with a new finite sum inequality. IET
Control Theory Appl. 6(3), 448–453 (2012)
35. Zhang, C., He, Y., Jiang, L., Wu, M.: An improved summation inequality to discrete-time systems with time-varying
delay. Automatica 74, 10–15 (2016)
36. Zhu, Q., Xu, J., Huang, D., Hu, G.: Iterative learning control design for linear discrete-time systems with multiple
high-order internal models. Automatica 62, 65–76 (2015)
37. Xiang, W.: Parameter-memorized Lyapunov functions for discrete-time systems with time-varying parametric
uncertainties. Automatica 87, 450–454 (2018)
38. Zhang, S., Meng, X., Zhang, T.: Dynamics analysis and numerical simulations of a stochastic non-autonomous
predator–prey system with impulsive effects. Nonlinear Anal. Hybrid Syst. 26, 19–37 (2017)
39. Meng, X., Zhao, S., Feng, T., Zhang, T.: Dynamics of a novel nonlinear stochastic SIS epidemic model with double
epidemic hypothesis. J. Math. Anal. Appl. 433(1), 227–242 (2016)
40. Zhang, J., Peng, C., Zheng, M.: Improved results for linear discrete-time systems with an interval time-varying input
delay. Int. J. Syst. Sci. 47(2), 492–499 (2016)
41. Chen, J., Park, J., Xu, S.: Stability analysis of discrete-time neural networks with an interval-like time-varying delay.
Neurocomputing 329, 248–254 (2019)
42. Shan, Y., Zhong, S., Cui, J., Hou, L., Li, Y.: Improved criteria of delay-dependent stability for discrete-time neural
networks with leakage delay. Neurocomputing 266, 409–419 (2017)
43. Kwon, O., Park, M., Park, J., Lee, S., Cha, E.: Stability and stabilization for discrete-time systems with time-varying delays
via augmented Lyapunov–Krasovskii functional. J. Franklin Inst. 350, 521–540 (2013)
44. Meng, X., Li, F., Gao, S.: Global analysis and numerical simulations of a novel stochastic eco-epidemiological model
with time delay. Appl. Math. Comput. 339, 701–726 (2018)
45. Meng, X., Wang, L., Zhang, T.: Global dynamics analysis of a nonlinear impulsive stochastic chemostat system in a
polluted environment. J. Appl. Anal. Comput. 6(3), 865–875 (2016)
46. Zhang, T., Zhang, T., Meng, X.: Stability analysis of a chemostat model with maintenance energy. Appl. Math. Lett. 68,
1–7 (2017)
47. Chang, Z., Meng, X., Zhang, T.: A new way of investigating the asymptotic behaviour of a stochastic SIS system with
multiplicative noise. Appl. Math. Lett. 87, 80–86 (2019)