Definition 2 [1]. A function b : R
þ
R
þ
!R
þ
is called a KL-function if for each fixed t P 0, the function bðs; tÞ is a K-func-
tion, and for each fixed s P 0 it is decreasing to zero on t as t !1.
Definition 3. System (1) is said to be stochastic input-to-state stable (SISS), if there exist functions b 2 KL and
c
2 K such
that for any bounded input u 2 L
1
and initial data x
0
2 R
n
,
e
ð xkkÞ< bð x
0
kk; tÞþ
c
ð ujjÞ;
8
t P 0; xð0Þ¼x
0
Remark 1. When u ¼ 0, SISS is equivalent to global asymptotic stability in probability by Definition 3.
Definition 4 [2]. Consider system (1). A continuously differentiable function Vðx; tÞ2#
2;1
ðR
n
R
þ
; R
þ
Þ is called a SISS-
Lyapunov function, if there exist functions
a
1
;
a
2
;
a
3
;
a
4
2 K
1
, such that
a
1
ð x
kkÞ
6 Vðx; tÞ 6
a
2
ð x
kkÞ
ð2Þ
LV 6
a
3
ð xkkÞþ
a
4
ð ujjÞ ð3Þ
for any x 2 R
n
and u 2 L
1
, where
LVðt; xÞ¼V
t
ðt; xÞþV
x
ðt; xÞf þ
1
2
tr h
T
V
xx
ðt; xÞh
hi
f ¼ Axðt ÞþA
d
xðt
s
ÞþB
1
uðtÞþE
1
gðxðtÞÞ
h ¼ DxðtÞþE
2
xðt
s
Þ
Remark 2. Another definition of the SISS-Lyapunov function is that the function Vðx; tÞ2#
2;1
ðR
n
R
þ
; R
þ
Þ satisfies the con-
dition (2), and the following inequality
LVðx; tÞ 6
a
ð x
kkÞ
;
8
x
kk
P
a
3
ð u
jjÞ
ð4Þ
holds, where functions
a
3
;
a
2 K.
Lemma 1. The system (1) is SISS if there exist a SISS-Lyapunov function.
Proof. See Appendix.
Here, consider a dynamical filter for estimation of the signal zðtÞ
dx
f
ðtÞ¼ðA
f
x
f
ðtÞþE
f
gðx
f
ÞÞdt þ B
f
dyðt Þ
z
f
ðtÞ¼C
f
x
f
ðtÞ
ð5Þ
where x
f
ðtÞ2R
n
is the filter state and A
f
; E
f
; B
f
; C
f
are appropriately dimensioned filter matrices to be determined.Define
nðtÞ¼
x
T
ðtÞ x
T
f
ðtÞ
hi
T
;
zðt Þ¼zðtÞz
f
ðtÞ; gðnðtÞÞ ¼
g
T
ðxðtÞÞ g
T
ðx
f
ðtÞÞ
T
Then, the filtering error system from the systems (1) and (5) is described by
dnðt Þ¼½
AnðtÞþ
A
d
N nðt
s
Þþ
B
1
uðtÞþ
E
1
gðnðtÞÞdt þ½
DNnðtÞþ
E
2
N nðt
s
Þdw
z ¼
LnðtÞþ
L
d
N nðt
s
Þ
ð6Þ
where
A ¼
A 0
B
f
CA
f
;
A
d
¼
A
d
0
;
B
1
¼
B
1
B
f
C
d
;
E
1
¼
E
1
0
0 E
f
;
D ¼
D
0
E
2
¼
E
2
0
;
L ¼
L C
f
½;
L
d
¼ L
d
; N ¼ I 0½
Definition 5. Given a scalar
c
> 0, the filtering error system (6) is SISS and satisfies a prescribed H
1
attenuation level
c
if the
system (6) is SISS, and under zero initial condition the following inequality
674 F. Zhao et al. / Applied Mathematics and Computation 227 (2014) 672–686